Skype, SIP, the Enterprise and Security
Transcription
Skype, SIP, the Enterprise and Security
Accelerating Growth of Skype Marks New Approach to Voice Communication Meanwhile Enterprise VoIP Grows but Requirements Very Different from Single User, Small Business World of Skype Adoption Is Skype Still VoIP? The first generation VoIP that has generated so much buzz in the last several years has been based on wrapping voice in IP packets and sending it through IP networks. Its advocates claimed that as it’s just a bucket of bits it had the potential to hollow out the PSTN and pull the phone companies over the edge of solvency. The early movers generated lots of buzz and spawned a sizable industry including companies that exist to enable their customers to use their present day phones to convert to VoIP in order to connect to the PSTN almost anywhere in North America and even in Europe for one low monthly rate. The process, although it uses protocols like H.323, Megaco and SIP (Session Initiation Protocol) to establish and route calls and to connect myriads of different devices within enterprise and carrier networks is still tied more closely than some would like to people’s expectations of the behavior of century old POTS service. The vision, enabled mainly by SIP, was to allow users to do many new and interesting things including to show presence but for reasons debated in our Symposium discussions, this vision has not yet been well realized. In any case, despite all the “buzz” related to VoIP, so large are phone company revenues and so small has been the potential uptake on these services that the phone companies’ bottom lines are still largely unaffected. In the late summer of 2003 a very different approach to VoIP arrived as Niklas Zennstrom, the young geek who in the preceding few years had built Kazaa into an extremely successful peer-to-peer file sharing program launched Skype. Skype was a free peer-to-peer, voice over Internet protocol, soft phone with the Global IP Sound codec that offered users voice quality far superior to POTS and would work effectively over bandwidth that was significantly lower than other VoIP devices demanded. Skype gave its product away - seeking to establish value in other ways. It worked computer-to-computer utilizing the resources of always-on broadband Internet to take some of the resources of its user’s machines and turn the borrowed resources into the end-to-end internet equivalent of telco central offices. This enabled free voice communication from any Internet-connected machine of even Wi-Fi or cell phone to any other Skype running platform elsewhere in the globe. Voice here was not only a bucket of bits, it also never touched the PSTN - unless a user bought the SkypeOut service. Starting not surprisingly on Windows, Skype moved over the next year to releases on Linux, Mac OS X and Pocket PC. It is also looking at possible releases for Windows SmartPhone, Symbian and Palm. In early November 2004 it was seeing 80,000 new downloads a day and hit its first million simultaneous users on line. By February 2005 it had reached 150,000 downloads per day and Volume XIV, No. 2-3 May - June 2005 ISSN 1071 - 6327 for the first time two million simultaneous users. As this is written on March 1, Skype shows 78,750,000 down loads. Skype Voice mail exists in Beta test. Skype video cannot be far behind. Skype’s architecture, interesting as it is, is not the last word. Still, I maintain that you cannot easily see what is significantly different about Skype unless you use it. I have been using it and talking to others who do the same. Skype is changing the nature of what it means to be online. In doing so, it is changing my assessment of the importance of broadband and what it makes possible. Broadband now is not just faster email and Web, it can also be used as an enabler of new ways of real-time remote collaboration. In early December on the COOK Report Symposium on Network Economics mail list, in the midst of a casual discussion of Skype, David Reed put his finger on some of the differences between the two different approaches On the Inside Skype and VoIP Contents p. 2 Please READ Explanatory Note page 6 With Arcobat reader click on blue text Contents Accelerating Growth of Skype Marks New Approach to Voice Communication - Meanwhile Enterprise VoIP Grows but Requirements Very Different from Single User, Small Business World of Skype Adoption p. 1 Is VoIP No Longer an Appropriate Term for Thinking About the Kind of Communication Represented by Skype? Some Further Points of View on the Meaning and Role of Skype p. 9 Interview An Introduction to the World of Skype Skype Specialist Stuart Henshall Evaluates Skype’s Strategy of Moving Voice and Real Time Collaboration to Wherever Broadband is Present p. 13 Symposium Discussion December 1 - January 4, 2005 Sype versus SIP: a Debate Between the “It Just Works” Point of View and the Standards Based, Interoperability World View p. 23 Dave Hughes Discovers Skype p. 23 Why SIP Has Failed p. 24 Is Skype a Proprietary System Creating a Wintel-like Platform? p. 27 Some SIP Issues p. 29 Platforms: Skype and SIP - Islands versus Standards-Based p. 30 SIP Must Interoperate While Skype Need not Do So p. 32 What to Do About NATs and Protocol Problems? p. 33 The Value of Interconnected Networks p. 34 What the Technology Must Do in Order to Please the User p. 35 Changing Expectations of What People Want from Telephony p. 37 A Changing Role for Operators p. 39 Symposium Discussion January 4 -15, 2005 Some Broader Issues Involving the Bluring of Boundaries of VoIP in the Telephony Wireless and Enterprise Worlds A Balkanization of Personal and Enterprise Communication Trends p. 40 Multiple VoIP Markets p. 41 Skype Interconnectivity p. 42 Meshed Up Neighborhoods p. 43 Skype Marches On p. 45 2 The COOK Report on Internet May June 2005 Access to Broadband p. 45 Viral Communications p. 46 “Harmonizing” Discordant Systems p. 47 Intel Backs Municipal Broadband p. 48 The Window of Opportunity for ENUM is Closing Fast p. 49 Two Parts of VoIP p. 51 Competition and Wireless Paths - WiMAX Troubles p. 51 Impact of Smart Design on Wireless Cost p. 53 How to Think about the Strategy of Power in Terms of Part 15 p. 53 An Almost Hopeless Complexity of Variables Involved in Service Pricing p. 55 Interview Skype Seen as “Instant Voice-Integration” of Multiple Forms of Communication into Broadband Based Collaboration -- Improved Audio Codecs, P2P Architecture, & Other Features May Push Skype like “Instant Voice” Softphones into New Areas --James Enck Explores Possible Impact on Wireline and Mobile Carriers Symposium Discussion Jan 15 to Feb 6 How VoIP Mixes with Wireless, the Enterprise and other Markets p. 57 VoIP Peering Architectures - Hard Installations or More Flexible Software Glue? p. 66 Strange Concepts of Free Markets in the Muni Network World p. 69 Google Wants Dark Fiber p. 71 Google as an Example of When to Route Versus When to Switch p. 73 Google VoIP p. 74 AT&T Looks Beyond “Number, Please” p. 77 Issues of Security p. 79 Commercial VoIP Pricing Power p. 79 Wireless Business Models p. 80 VoIP Pricing Power Encore p. 82 VoIP and Security in Government and the Enterprise p. 84 Symposium Discussion Feb 6 - 22 Skype, SIP, the Enterprise and Security 3 Skype is Like Apple II in the Enterprise p. 88 Is the Enterprise is in Gridlock? p. 89 Stuart Henshall and Skype Voice Mail p. 91 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Successful Video Accommodation by Microsoft’s p. 126 IPT Platform Cisco’s CMX Platform Promotes Mobility Across Wireless Access Networks p. 93 But What is the VoIP Market?, p. 94 The Importance of Directory Services, p. 96 Skype and Motorola Marketing Partnership, p. 97 Some VoIP Regulatory Issues Goroshevsky, Popular Telephony and Peerio p. 98 Vonage Whining its Way to Open Access, p. 129 How to Explain Why Skype Works Better? p. 133 Cringly Asks Have Best Days of VoIP Come and Gone? p. 133 Voice, Caspian, Broadband and Korea, p. 127 IMS in the Mobile World p. 128 Symposium Discussion March 3 - 7 Enterprise Voice Issues Yield Many UnKnowns p. 99 Skype Not Tied to Specific Hardware Can Act as a Virally Infectious Communications Agent p. 101 Skype from an Enterprise Point of View p. 102 But Are P2P Voice Applications Blockable in the p. 136 Same Way as Vonage or Lingo? Skype and Enterprise Security Issues p. 103 What the Regulator Will and Will Not Do, Skype, Firewalls and Security p. 105 Vonage Suffers Widespread Outage on March 4, p. 139 Skype in the “Civilian World” p. 107 Skype in Hotspots, But Skype May Show up in Some Enterprises Sooner Rather Than Later p. 108 Symposium Conclusion March 11-17 Skype, Web Services and Mission Criticality, p. 109 Sip Based Enum Wi-fi Phones p. 110 On Walled Gardens and Getting to the Other Side of Geoffrey Moore’s Chasm p. 111 Skype and Grid Computing p. 112 The Google Telephone Network and a World of Abundance in Communication, Skype’s Impact on Voice Traffic and Open Source VoIP PBXs, p. 113 p. 114 p. 117 With Regard to Architecture and Economics all p. 118 VoIP is not Alike, p. 119 QoS OpEx Economics - Flows or UCLP? p. 122 Technical Aspects of VoIP Traffic Shaping on Wi-fi Network p. 124 Rethinking Skype Mega Chats, p. 144 Choosing How to Connect with Each Other Becomes Critical, p. 146 Skype In Looks to Be an Important Intersection Between Skype and SIP, p. 147 Highlights p. 151 Executive Summary p. 180 Side Bars VoIP Economic, Quality and Network Traffic Issues More on Muni Networks (Texas and Colorado) p. 142 Instant Voice, Chat and Messaging Motivate Early Adaptors to Rethink What it Means to Be Connected p. 143 Symposium Discussion Feb 25 - March 3 VoIP Adoption Curves, p. 137 New York's Broadband Gap p. 68 Millimeter wireless Links p. 75 Peer to Peer SIP p. 75 New Push for Wireless p. 76 CODECs and Perceptions of Voice Quality p. 131 A refutation of Metcalfe’s Law and a better estimate for the value of networks and network p. 141 interconnections, by Andrew Odlyzko 4 The COOK Report on Internet May June 2005 Symposium & Interview Contributors to this Issue Affilation given for purposes of identification - views expressed are those of the contributors alone Nigel Ballard, Wireless Director, Matrix Networks, proprietor of joejava.com Malcolm Matson,British entrepreneur and Director of OPLAN Foundation Jim Baller, Partner in Baller Herbst law firm & Expert on Municiple Networks Franscois Menard, Canadian policy expert and municipal fiber network architect Sebastian Buettrich is a technology strategist who founded wire.less.dk in 2002 Andrew Odlyzko, Director Digital Technology Center, University of Minnesota Mike Cheponis, wireless consultant, antenna design specialist Dave OʼLeary, Juniper Networks Frank Coluccio, President DTI Consulting, NYC, high-capacity optical netwʼk consultant David Reed, Internet pioneer, spectrum policy expert, currently with Media Lab & HP Peter Cohen, consultant and peering specialist for Telia Jere Retzer, Sr Mgr, Next Generation Networks, Oregon Health & Science University Melissa Davis, optical network architect with RS Information Systems Larry Roberts, Arpanet Pioneer, CEO Anagran Bill St Arnaud,Director Ca*Net4 , Canarie, Canada Peter Ecclesine, Market Analyst for Wireless, Cisco David Sandel, CTO, NetLabs LLC, St Louis James Enck, Securities Analyst & proprietor of Eurotelcoblog Chris Savage, elecom attorney and partner at Cole, Raywid & Braverman in Wash. DC Jim Forster, Distinguished Engineer, Cisco Martin Geddes, consultant and author Telepocalypse Henning Schulzrinne, Professor and Chair in the Dept. of Computer Science, Columbia University Vijay Gill, Director Peering, America on Line Ron Sege, CEO Tropos Alex Goldman, Editor ISP Planet Jupiter Media Raj Sharma, President of Nextone a VoIP systems integrator Steve Heap, CTO of Arbinet a bandwidth broker and VoIP Traffic terminator Richard Stastny, Austrian Telecom author VOIP and ENUM Blog Sebastian Hassenger WebSphere Market Strategy and Planning IBM Software Group Richard Shockey, Senior Manager, Strategic Technology Initiatives, NeuStar Inc. Stuart Henshall, consultant and author of Unbound Spiral and Skype Journal Jim Southworh, former chair DSL forum, VP Concentric. Now Secure Pathways CEO Tom Hertz, CTO, Opportunity Iowa Jeff Sterling, Interconnected Associates, Bellevue, Washington Dave Hughes, owner Old Colorado City Communications and wireless advocate Matt Wenger, Product Manager North America, PacketFront Cullen Jennings, SIP & VoIP Security Expert Cisco Damien Wetzel, Network Consultant Paris, Formerly with Akamai and Internap Patrick Leary, Wireless Evangelist, Alvarion Ron Yokubaitis, CEO Giganews Tony Li, Router Architect at Cisco, Juniper and Procket, recently returned to Cisco 5 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA to VoIP when he stated: “The way to succeed in business is to pick the best customers, and delight them. And the crucial caveat - the best customers are not the ones who always buy anything you sell - those are your best customers, not the best customers. The best customers are the ones who will teach you what you should be selling.” “The following is how it applies here: “SIP’s vendors have defined their customers to be phone companies. “Skype has defined its customers to be people who live a communications-centered life. “It’s impossible to delight a phone company with voice over the Internet. The people who live a communications-centered life will teach you what really matters. Those people are *not* happy customers of the phone company.” David’s statement is a rather sweeping generalization and one that is not entirely fair to players involved who were constrained in what they could do by virtue of the fact that any approach to complex enterprise phone systems would face restrictions that an unconstrained greenfield approach like Skype would not. Still a sweeping generalization may be useful if it captures the stark contrast between the two worlds where user expectations are very different. Something very interesting is going on here. This two month issue of the COOK Report explores that by looking at the early success of Skype in the larger context of more traditional VoIP services. While Skype is not the be-all and endall of approaches to VoIP, it does open many new possibilities, even though, for reasons that we will explore later, it will not easily find its way into large-scale corporate use. Getting VoIP Adoption that Scales In mid-December on our Symposium list Reed exclaimed: “Sending voice over IP is trivial. That’s not the technical problem. Getting scaled adoption is hard, and a common standard that works simply was required. SIP could have been a contender. It isn’t going be. And I think its own ‘proponents’ killed it. “SIP should have won, because it is an open standard, but the desire to create a business model that captures the old unsustainable voice revenues of the RBOCs has seduced Cisco and its customers into waiting and making the standard more complex. Unlike the old days of the Internet, where interoperability was the centerpiece, the likelihood that a SIP phone will work with one from another vendor is near zero. There was a reason that the major IP trade show in the early days was called ‘Interop’!” David’s assertions here are certainly inflammatory to some. Can one know for certain that Cisco allegedly ruined SIP by selling out to the RBOCs desire for selfpreservation? Very unlikely. I claim no inside information, but these companies have been huge customers of Cisco, Lucent, Nortel, and a whole range of other equipment suppliers. That they would listen to their RBOC customers would not be surprising. Nevertheless, there is also evidence that they are improving and that lack of interoperability is no longer nearly as bad as David states. The point is that we are in the midst of major disruptive change to the voice service that has sustained the phone company’s bottom lines. While struggles should not be surprising, we should also recall that the process of making a kaleidoscopic array of enterprise telephony equipment interoperate with new protocols designed to get them to do new things is a demand- ing task as well. Meanwhile Adobe, in the mid 1990s, wanted to create a standard document format exchange program. The goal was to dump a document in and have it come out in a universally readable format. It created Acrobat and then gave away the program necessary to read Acrobat documents. It succeeded and today the page definition or PDF file has become the nearly universally accepted standard. Riding on its early success, will Skype do the same? With nearly 80 million downloads in its first 18 months, the question is whether Skype has so well-seeded its user base that a better program cannot overcome it? Perhaps. Perhaps not. However, there is one thing that it is likely to prove: telephony now really has become software. And in becoming software it can be separated from the physical infrastructure of the old phone system far more easily as it rides in its new form on broadband IP inter-networks. Of course there are still many hundreds of million phones in use connected by twisted pair and other means and making up what we call the global switched telephone network, or locally the PSTN. As broadband IP networks grow they represent a fertile field onto which the traffic from the PSTN will switch. Of course, how rapidly this will happen is still a matter of conjecture. Many, many problems remain to be solved. Viral Communications and Skype The communications industry is facing disruption. The technology giants - in this case the local phone companies- a decade after the internet went commercial, are running outmoded centralized infrastructures with revenues of hundred of billions of dollars per-year . While they are changing their infrastructure, whether they are doing it rapidly enough An Important Note to Our Subscribers - Blue URLs are hotlinked with Acrobat -click them This is a two part May - June 2005 issue. Please start with the Exec Summary on page 180. Then read the two introductory articles beginning one page one and nine. Next please read the two interviews (pp. 13 and 55). After that either plunge into the Symposium Dissussion or go to the highlights section beginning page 150 and use the blue page links to dip into the Symposium on an as needed basis. The next issue WILL be shorter. The first 54 pages were originally published on March 6. 6 The COOK Report on Internet May June 2005 is often debated. I will posit here that continuing advances in hardware and software design are creating for the first decade of the 21st century what Reed and Lippman call “A Viral Communications Architecture.” [See the May 19 2003 paper at http://dl.media.mit.edu/ viral/viral.pdf.] While this paper focuses on wireless and software defined radio, I contend that its most central premises also fit Skype. The Reed -Lippman argument defines a viral communications architecture as follows. A viral architecture “is one where elements are independent, scalable and where each new element adds capacity to the system, so that it can be adopted incrementally from a small base and gains accelerating value with scale.” (p. 2) “The key idea is [the emergence of] communications devices that work with no central backbone and scale almost without bound.” (p. 2) Looking a recent technology history, the authors point out that in this environment change is enabled by “the vastly reduced cost (or risk) of innovation. Barriers such as high integration costs, centralized support and a brittle architecture don’t exist in the PC. The cost of change is borne incrementally, as each user purchases a new package. Software can become a garage industry and a growth potential can be detected earlier.” (p. 3) The user can purchase the assets and tools needed for his own production and put them to work in a business venture that he initiates and in which he invests. I contend that as went the PC, so goes the new tools of viral voice communication. However, in fitting viral voice communications into the Reed-Lippman viral wireless bed, there is one critical change and that is that the Skype voice software is dependent on the broadband infrastructure laid down by the telcos and cable cos in the first decade of the commercial Internet. While depending on this physical infrastructure that the old industry laid down, the new voice software is spreading parasitically and virally. Although Reed and Lippman focus primarily on viral software-defined radio that is largely free of the need to use physical infrastructure, the viral VoIP software that we explore absolutely depends on the broadband infrastructure laid down by the companies that it potentially obsoletes. This will create tension but, as that infrastructure is used by virtually the entire spectrum of so-called normal communications and since its peer-to-peer architecture makes it very difficult to discriminate against. It is hard to see a scenario in which this viral VoIP could be denied its carrier. I further contend that with Skype we have a phenomenon that matches the requirements laid out by Reed and Lippman when they state that: “Each new element of a viral architecture must not deplete the capabilities of those that were there before -- to gain momentum, each new element must create more value from connecting into the system than from operating alone. That is, each adoption is a “win-win” decision - the existing elements gain a little more benefit from the new element, and each new element has a stronger value proposition for joining the system. Momentum results from this process, because a reluctant adopter will eventually be attracted to adopt when the scale reaches his cost/benefit tradeoff even when the architecture still has small reach. In the case of fax machines, this happened when enough of your contacts had or used fax machines that the case for owning one became compelling. A virtuous cycle results from a growing market cutting manufacturing costs, and increasing benefits to each new purchaser.” (p.5) And since Skype is free, the only investment needed to begin use it is some time to install and someone with whom we wish to communicate where the resulting change in communication modality will become cost effective. “There are two primary design principles that lead to a viral architecture: scalability and independence. The first states that a viral system ought to be able to grow almost without bound, and the second requires that its elements operate autonomously, without connection to a central authority. In essence, one should be able to freely add elements and they should work without connection to a backbone.” (p. 5) The peer-to-peer nature of Skype, as software that is largely platform independent, enables it to nicely meet these 7 requirements. “In the end, viral communications transforms communication from something you buy to something you do. Independence of operation allows communications services to be separated from traditional service providers. At one [very fundamental] level, this is a pure threat to the extent that the economics of the telecommunications industry business model depends on exclusivity of service provisioning.” (p. 10) These paragraphs I believe are a good illustration of what Skype represents. It is a form of VoIP vastly different from what has come before. It is not the be-all and end-all of this form of development and indeed, as our discussion will show, its nature as a viral carrier will leave it profoundly unwelcome in many areas where security and control of communications is paramount. In the pages of this issue we will also explore in depth what is being done with the more standard VoIP applications that come in forms that the phone companies and the large enterprises are more used to and can much more easily control since they are mapped much more closely to the standard telephone networks. Even with Skype, the PSTN won’t vanish over night. There are hundreds of millions of people without the necessary computers and hundreds of millions more without access to the necessary broadband. The change of mindset required to look at your computer as also a telephone will take some time. But I now believe that the flow of change is faster than I assumed only four months ago when I first used Skype. Before the first generation of VoIP had fully settled in we are being swept by a second generation. A lot of people, including quite a few on our symposium list, are raising doubts about Skype. Many are justified. As eloquent as the Reed-Lippman viral point of view of the universe is, the business model still isn’t clear. For, if communication becomes what you do rather than a service you buy, we must still buy the needed hardware and find a way to enable maintenance of the basic infrastructure on which the Internet depends. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Still in the Early Stages? On March 1, Reed wrote: “I’m . . . optimistic that dominoes are falling. Cellular is destroying wireline, LD ‘carriers’ are pretty much dead on their feet and surviving by providing commodity pipes for data and voice-over-data. IP is already in use in the telephony infrastructure (PTT except on Nextel’s legacy network is IP-based). Presence and IM are growing markets. File sharing is being monetized by Apple and its friends, and being experimented with by BBC and others.” Frank Coluccio added: “Also Ethernet in the metro and WAN, particularly when the Enterprise owns its own fiber, is neutralizing, where it is not also a tenant of, the SONET that came before it, allowing enterprises to self-provision their own network topologies and configurations in situations where previously they were fully dependent on carriers.” Reed continued: “I love what Skype is doing, but it’s currently the Apple ][ of the new end-to-end personal communications experience. With luck they’ll make it to the Macintosh level. No one has legitimized personal communications as IBM legitimized the PC, yet ... But Motorola’s deal with Skype indicates what I see in my conversations with the *non*-marketing people at many communications companies - they see the future and are willing to take risks, albeit tiny ones, that may not be enough to become players in the next wave. . .” “[Meanwhile] the marketroids are playing their usual anti-novelty role, spinning fear, uncertainty and doubt at their consumer and enterprise customers. (No one really wants a GUI. It’s inefficient. Why would anyone who does word processing want a PC with a color screen?) But in the home offices, away from the Wall St. analysts, they know that isn’t going to fly for long. That said, it’s not time to fall in love with Skype, or the iPod, or any short-term vestige of a possible future. They are merely indicators, and, if they are to benefit from what they started, it will take hard work,” Reed concludes. This is a two month issue that looks at both Skype and at more traditional use of VoIP in carrier and enterprise networks. It will show that what most enterprises want may well not be Skype. But while doing that it will explore the potential for the viral growth of Skype. In some cases Skype may become an infectious carrier that enables PSTN or cellular usage or even dial-out to PSTN usage to be avoided. GSN and 3G smartphones are predicated on getting users to pay for convenient access to Internet, email and SMS messages. A major question is whether Skype will free its users as WiFi spreads from needing to pay the cell phone companies? Skype changes fundamentally what it means to use a phone by becoming the central voice application around which a layer of written search and information exchange tools are built. It enables realtime remote collaboration better than anything else so far developed. With voice mail, it greatly facilitates the exchange of audio messages. With FTP file exchange services, it enables groups of conference callers to share and comment on documents charts and pictures. Talking faces is the only thing missing from real time video conferencing benefits. I must point out that dividing what is happening into two generations of VoIP may not be entirely fair. The enterprise user base and requirements for the first generation of VoIP are so vastly different from the base being served by Skype as to make it difficult to talk about both in the same breath. VoIP in the enterprise faces vast levels of complexity that Skype riding peer-topeer on the public IP internet does not. In the enterprise, VoIP must deal with the need to do many things that Skype’s architecture enables it to avoid. Enterprise VoIP must cross system boundaries at levels that include many different protocols, addressing mechanisms, requirements of travel off and on the PSTN - transition through public IP networks and private VPNs - travel through firewalls with a myriad of attendant security issues and the need for inter-operability of equipment - some of which may be made more for carriers than enterprises. H.323 to SIP conversions and different implementations of SIP proxies in corpo8 rate DMZ boundaries further complicate matters. The result is spaghetti mess that one need not blame on SIP or really on anyone thing besides corporate need for security and the fact that enterprises are bound to operate in the real world of equipment much of which was put in place in a time of rapid technological change and therefore can hardly be expected to inter-operate smoothly or be replaced over night. We find that Skype’s proprietary architecture could be dangerous if it were hacked to enable worms and other kinds of malware. One should remember that security with Skype goes well beyond encryption of the conversation to the Enterprise IT systems level where security means something more than just encrypted conversation. Cautionary flags are called for. However one should also remember that enterprises are also infiltrated top to bottom with other proprietary software (especially that from Microsoft) that enables serious harm when security is breached. Some of our experts express profound misgivings about Skype. We give them ample opportunity to express their doubts in the long discussion that, apart from the interviews, makes up the body of this issue. Nevertheless, as we watch what Google is doing in its local service and other new features at the same time as Skype spreads, we wonder if we are not looking at a wave of approaching change as important as that of the maturation of the World Wide Web a decade ago. A final note - this issue is focused much more on Skype and its disruptive impact on voice communications outside the large enterprise where it is migrating from fixed to wireless and portable connections. For reasons that we will make clear, the innovation and opportunities look to be here rather than in standard enterprise-based VoIP, about which we, nevertheless, will have a lot to say Editor's Note: We offer our thanks to Ed Ciesla - eciesla@sprynet.com - who in addition to an outstanding knowledge of telecom is a very very good editor and has again worked long hours on this issue. Is VoIP No Longer an Appropriate Term for Thinking About the Kind of Communication Represented by Skype? Some Further Points of View on the Meaning and Role of Skype Editor’s Note: The article that begins this issue went though some considerable revision as the result of comments on the Symposium mail list. Feedback was that I had too uncritcally adopted David Reed’s skewering of SIP. Consequently I tried putting the general drift of Davids critique into my own words. COOK Report: David pointed out that his critique was business and not technical - but still he is using technical terms like SIP as a protocol to make his point. – As a result some you all are still displeased. It seems to me the key difference that we are talking about is that you “engineers,” nearly a decade ago (I have been hearing this since I interviewed Rich Shockey on his fax protocol I think in 1996), began developing IP technology to replace various parts of the traditional telephone networks. Lord knows I am not a historian of SIP but I think I was interviewing on it back in 1999 - maybe even 98 with Henry Sinnreich explaining why it would knock brain-dead H323 out of the water, I know in the 1998-99 time frame I interviewed the developers of Megaco, and Henning in September of 2000. So these various protocols have been around a good long while and I would submit that when they first came to pass, they were folks working on ways to replace the more traditional telco networks, or shall we say bring them into the IP age? The battle cry I heard from the beginning was that this would hollow out and suck dry the PSTN. Now had anyone done a peer-to-peer application before 2001? Let alone a voice peer-to-peer application? I don’t think so. Could it be said that when Henning and colleagues began their work the customers were mainly telcos? I think that it’s reasonable. But could it also be said that they had ideas about how to do really cool things that never would have occurred to a telco? You bet. Could it be that David’s point is that what we have here is a collision between a huge group of very talented people whose early customers were phone companies. [Frank Coluccio points out below that the early customer were reallt would be competitors to phonecompanies.] While their customers now are new companies that they helped to design and build so that as the market evolved they could continue to sell VoIP boxes and services to enterprises and phone companies? I think those are reasonable assumptions. Frank Coluccio: Well Gordon - Not quite. As someone who was eyeball-deep in the creation of an Internet Telephony Service Provider venture during the late Nineties, I’d say that, with a few rare exceptions, VoIP developers of all types, including SIP developers at that time, were not targeting Telcos, per se. That’s where they wound up, but that was not paramount in the original game plan. Instead, they were targeting ITSPs, ISPs, international settlement houses and operators like iBasis and ITXC, greefield DLECs/CLECs, and of course, enterprises, for use over intranets in order to replace first and second generation IP to TDM and PSTN TDM to IP gateways, all of whom were targeting early-stage arbitrage opportunities. The telcos came along begrudgingly, later on by a couple of years, but not before SS7 interworking was finally worked into the framework, first. Melissa Davis: Engineers build to requirements, build for function, rarely assume that there is only one way to solve a problem, and work assuming “successive approximations” with the expectation of surprise. Smart Engineers know that the 9 success of a project at a developmental stage changes the requirements as adoption takes place and novelty follows, requiring redesign. SIP was an engineering project, not different from most of the IETF projects in this way. I can validate what Frank is reporting, not because I was a SIP developer, but as a reviewer and tester and quality reference, having built packet voice as VoATM and H.323 (software). The target, at least at Cisco, was not the telcos, but the Enterprises, toll-bypass, and selling where Cisco had the branding and adoption advantage (Lucent and Nortel had a lock on the telco voice switches, and Lucent’s demise was its “buyingtime” continued investment and revenue projections on its Class 5 switches and its “Broadband Manager”). COOK Report (continuing last night's summation): So what happened? Some uppity young geeks who built a very successful file sharing network for pop music (Kazaa ) said Hey let's do a peer to peer program to share voice. The result was Skype. It was a greenfield approach that was unencumbered by anything and everything having to do with 100 years of telco history and infrastructure. Being unencumbered allowed them to put all manner of stuff into their voice sharing that those of us who had not tried it would never have associated with making telephone calls. Now to be fair, the SIP community saw that they could design voice communication systems that would use presence and instant messenging and those kinds of tools in very innovative ways. They were well aware, I suspect, of the possibility for delivering all the other kinds of benefits brought by Skype. Freed from all outside constrains, they could have delivered better than Skype. If they had been able to behave in the really greenfield way as could and did COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA the developers of Skype, then we would be seeing more Skype like products out there. Instead they got tied up, tied down, emeshed - or whatever you wish to call it. David’s characterization is quite blunt but I think in a broadbrush way its accurate. He who has the gold makes the rules -- and the folks with the gold had tons of teleco networks and equipment that they simply could not throw out over night. The result was an accomodation of that situation. It's no one's fault - but given the players and conditions, it was also not surprising. Skype was unencumbered and therefor advantaged. Skype has have taken advantage of its lack of encumbrance. The H.323, SIP, Megaco folk have performed magnificently given what they had to contend with, and the reason we are talking all these protocols rather than just simple VoIP is that - given the complex network contexts in which they had to work - it was only be means of these protocols that they could make voice packet instead of circuit switched. That is how I see in my own words the points that David was making. Meanwhile, Stuart Henshall chimed in on March 4: Gordon, It may be late to try and be provocative and take a contrarians view. Still I’ll try here. I’m obviously a Skype proponent, I’ve used it longer than most of the employees in Skype, and it matters little to me whether it is SIP or not. Oh I’ve keep digging to learn what SIP will do for me, and yet it still isn’t explained in terms that create desire and add new meaning to my life (see my intimacy example below). Still I wish Skype were open source etc etc. But that is just not the case. A few items, a contrarian’s points that may be worth exploring. 1. Skype spent the last year moving closer to the PSTN. With SkypeOut etc they created an interconnect that is satisfactory at best. Be careful not to accept too easily that Skype’s opportunity is to act like a revolutionary phone company. Rather it may be using this strategy to suck the lifeblood out of them. It stimulated a cost battle amongst the traditional incumbents. It destroyed international call rates, and is making the investment community “scared” of these same incumbents. Perhaps it forced the beginnings of a massive consolidation. All in all that helps Skype. For the new PoIP players they think they are still in the game. They are not, they are just cutting costs. The real caution is in the idea that “You can’t play a CD on a record player. Skype has been trying to - perhaps even encouraged to do this by investors. It’s short-term. But the direction for the product is well away from the PSTN. Look at the adoption of DVD’s. Look at global uptake of mobile phones. Skype is going fast after cheaper Wi-Fi enabled handset devices. By Xmas the Broadreach type exercise will only involve a $100 - $150 handset and it will be small! The biggest flaw may actually be Skype is based on GIPS. Unless GIPS provides a dynamic 3D audio codec Skype is vulnerable on sound quality. I hope for their sake they keep it in mind. Every Skyper who tries a 3D conference call will switch if it is offered for free. Similarly there are some “always” on features that will make a difference. Conclusion, connecting to the PSTN is temporary. 2. I’m concerned when I see the “enterprise” discussion. I realize there is a whole industry that is set up to deal with the Fortune 500 or 1000. You list is supported by them. But the fact remains that the majority of business is not actually done that way. The complexity that ties the mega’s up and the “security” approaches implied and referenced is not relevant to the way SME’s operate. Skype is already capturing independent global workgroups. I’m one, running elements of my business that would have broken my bank account two years ago. For the same reason that blogs and wikis are beating mega buck content management systems, Skype, and potentially other alternates are going to smash the traditional barriers. I’d really spend more time thinking about it in the perspective of SME’s. You can bet that Skype will use a blog like ap10 proach to get adoption in enterprises. You start with a project team. You don’t need approval for that. You just do it. Groups with a high element of travel will drive this faster and faster. I know Fortune 500’s that are already in this situation and similarly groups within “major” players in China. 3. The argument here is premised on the American/European/Japan? world of business. I’d look a lot deeper at China, India and ask yourself if you were organizing, creating a new mesh: what would be adopted and where and what infrastructure would you use? I have a suspicion that Wi-Fi handset could be very big. I’m afraid that the security issues above are predicated on a profile and understanding of desktops and laptops not cheap micro mobile devices. The global telecom replacement is likely to come out of China. We may be the late adopters. Users will also determine this standard. 4. I mentioned audio quality in passing above. This discussion should not be had without thinking about “presence” and “intimacy”. When you have listened to a Skyper (like I did this afternoon) tell me about how he was in his hotel room laptop on his bed talking to his partner (same setup) and telling me he could have left it on all night just so they could share breathing you are beginning to get to the reality. The phone can’t deliver this. An Economic Power Shift? 5. The real end game is in economics. When consumers have always-on connections and unlimited personal storage, then economic power shifts into their hands. Visa and eBay both model elements of what could emerge. Identity is moving to the fringe. It’s a powerful aspect. Skype’s long term role may be that of info market facilitator. Telecoms weren’t designed with user reputations systems in mind, swarms, smart mobs or cooperative behavior. Are we moving a step closer? I did say I needed a rant tonight. Thank The COOK Report on Internet May June 2005 you for allowing me the opportunity. I’ll continue to the merry end. The discussion is framed as VoIP. In fact it should be framed on the quality of how we converse, how well we are enabled to talk and listen. Think next “gen” / net “work”. It’s also late at night, I think VoIP sucks as a term. It’s no wonder Skype is becoming a verb. On Skype numbers there a couple that are a little off, not materially. I think the downloads number is not helpful. Current Skype accounts number some 24 million and they currently claim 785000 SkypeOut users. It still peaks at about two million concurrent users online per day.(Is that 3x Vonage? On what investment?). Skype also claims the marginal cost of obtaining a new user is 1 cent. There’s some fiction there but the point is effective. Vonage's new user cost is in the hundreds of dollars. SkypeMe, Stuart Melissa Davis: Stuart is correct above to reference that the enterprise arguments I have advanced are based on the Japan/ North American/European models. India and China do not have SEC, banking, privacy laws and compliance regulations. Labor rates are far lower and the proportion of manual labor much higher, thus the cost of outages due to cyber-vermin at least less of a risk. At this moment, among my customers, Skype is unacceptable, not because it is “Skype”, not for political reasons, not because some group wants to “carve out the guts of the telcos” (political speech inciting to pejorative emotional elicitation), but because of the way it works to the extent we can tell, and the interactions it must have to known other exploits (i.e., some applications on the host machine have to play the sounds). If David Reed and Gordon Cook wish to tear down the protective walls of the enterprises, and the enterprises do not want those walls torn down, then that is a war path. A Skype proxy may well solve that problem with no loss of life. For reasons Jim Forster and I brought up earlier, Henning’s statement below that enterprises will be late adopters is in all probability true. The cost of change is huge with any large footprint. The MBA’s require justifications and reviews. The Security organization requires risk assessment and mitigation. IT simply lacks the senior staff depth to do the lab tests. Vendor’s put off by our customers for this reason have come to us: “we will bring the hardware, and set up the systems.” I don’t have time either to write test plans, methodologies, etc. for all the applications of promised utility. What would be the reason to change for the enterprise? Certainly not expense. Voice is already close to free for these large purchasers. The worldviews of the enterprises are unlikely to be persuaded by your cheerleading for Skype, nor threatened by it. They build private nets for internal communications, and do customer contact in DMZ like safe zones. They don’t care about “meaning of your life”, or about “intimacy”. The Skype model is always on interconnectedness to everyone else, total interdependency. Such requires an enormous amount of trust, which is anathema to the competitive and regulated capitalistic economic system in which these large enterprises play. This isn’t to say that things won’t change, just that cultures change far more slowly than individuals, and individuals far more slowly than technologies. David Reed: Melissa’s quite right. For those in the Enterprise who buy the “enterprise religion” (the way of the technocrat) change will be slow, and if Gordon wants the “enterprise” people to pay for his newsletter, he might want to tell them what they want to hear. In my experience in business (nearly 20 years of my professional life, the rest being in research), every company contains an “enterprise” and a “business” side by side, and those people in the “enterprise” have a bad habit of never understanding that they are nearly invisible ghosts in a living business. These are the people who could see no value in the personal computer over 11 Wang word processors and IBM 3270’s accessing COBOL apps, no problem with the “application backlog” because an app wasn’t real unless written to specs by a systems analyst, no reason to switch from plastic slides on overhead projectors to PCs and projectors, and no reason to pay for employees to connect in from home. These are the people who say that we must keep our secrets locked in central repositories, and require manager’s signatures for every nickel spent, because empowering employees to make decisions is too risky. They are the legal dept., the CIO (if the company is unlucky) and the HR dept. - the people who think they make the business run, but really mostly contribute at the level of corporate hygeine at best. In fact, they make a fetish of saying no to change, and in the security area a fetish of confusing top-down control and the prerogatives of management with protecting the interests of the company. So this won’t happen. And it’s not just Skype, but ALL technology change that they resist. SIP, too. So it must feel really good to make small inroads into the “enterprise”, rather than selling to all the businesses on the planet. Doesn’t it seem like a paradox that Skype is more secure than ANY phone system out there, because it is end-toend encrypted, but that security is being turned against it by the very guardians of security who have deployed completely hackable systems in their companies? (I do fault Skype on not making their security architecture more transparent, but compared to SIP’s bolt-on-hodgepodge theory of security, I’ll take Skype’s until someone builds a the next great voice product by assembling a plug-and-playsecure-SIP-based product). That people who probably know that viruses spread like wildfire once they get inside their enterprise firewalls - due to basing their corporate architecture on boundary-based security - still object to deploying a new product that is inherently more protected from hacking, because it has to live in the dangerous COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA world outside the firewall? That p2p techniques that reduce the need for centrally managed servers are resisted by the group in enterprises whose budget directly depends on the number of servers they manage? I’ve watched in my career how disruptive technologies (email, spreadsheets, PCs, LANs, the Internet itself ..) invade the enterprise. They never come through the enterprise sales force, and they never come in the IT doorway. Their expo- 12 nential growth starts insignificantly and away from the center. But they transform the enterprise anyway. So Melissa’s right. “The enterprise” will never adopt this technology. But the real, pragmatic businesses will. An Introduction to the World of Skype Skype Specialist Stuart Henshall Evaluates Skype's Strategy of Moving Voice and Real Time Collaboration to Wherever Broadband is Present Highlights Editor’s Introduction: Stuart Henshall has been a management and marketing consultant and was affiliated with the Global Business Forum in the 1990s. He downloaded Skype on Day 3 of its release in the summer of 2003. His blog, Unbound Sprial, http://www. henshall.com/blog/ is devoted primarily to Skype. A November 8, 2004 interview with Niklas Zennström is found at http://www.engadget.com/ entry/2635319328796286/ I interviewed Stuart on February 18, 2005. COOK Report: What do you see then as the context and big picture of what is going on with Skype? Henshall: Let's start by considering the OS. When Skype originally launched it was as a Windows platform and there was a lot of discussion that said: “Oh Skype is just another windows application”. Folk concluded that they would never develop for other platforms because with Windows they had the whole world and therefore they didn’t have to. Consequently their first interesting strategic move was in going multiplatform - something that for a small software company was enormous an cost. But they went on to support Mac OSX and Linux. Concurrently with this they built a PDA version for Windows compatible PDAs. Consequently we are sitting here today with Skype being available on all major platforms with the exception of Palm - if you consider Palm to be major. You also have some evidence that says they are either looking at or doing work in the Symbian area. Multi Platform and Peer to Peer - A Force to Change the Basic Essence of Telephony COOK Report: Would you give me some context for Symbian? Henshall: Symbian came out of Psion. Psion was the operating system for one off the earliest PDAs that came out of Britain. But I guess originally it might even trace its life back to Sinclair. Symbian has managed to become an open source developer platform that supports 3GSM mobile phones. It is owned I think about 51% by Nokia and 49% by Motorola and some other parties. [Editor’s Note: http://www.symbian. com/ notes that shipments of Symbian OS phones doubled in 2004 to 14.4 million phones - the third consecutive year in which Symbian OS phone shipments grew more than 100%. Forty-one Symbian OS phones are currently shipping from eight Symbian OS licensees to more than 200 network operators worldwide.] The major cell phone manufacturers outside of China and Korean are now part of this little group devoted to Symbian as their OS. Smartphones are referred to as the category of cell phones that handle all the PDA kinds of things (email, web browsing and so on) as well as delivering voice. There are really only two OS platforms for smartphones like the Nokia 9500 Communicator which is currently the premier example of this technology. The other OS smartphone platform is what Treo and Windows compatible smartphones run on. Nokia still has at the moment the largest market share for smartphones at about 75%. 13 There are several things in the Symbian operating system at the moment that seem to leave it pretty well prepared for the future. One of the areas is audio and sound. It already has all the necessary capabilities in it for playing MP-3s, stereo music, or streaming music through it while the comparable Windows platform still has problems with that. I just saw an announcement from Japan for stereo audio and 3-D audio headsets for cell phones. The result is a dynamic soundstage much like you have in gaming environments or in your home theater. There is a lot of work being done in the sound and audio area, because quite clearly the iPods of the world are getting very close to converging with these things. When you have a PDA or one of these I-mates that is really just a Windows PDA, you insert a memory stick and all of a sudden you have a megabyte of music in your smartphone for an extra $50. COOK Report: It sounds to me that you are saying if they are going to be multiplatform, it is only logical for them to move onto to cell phone platforms including what they announced a day ago with Motorola? Henshall: Yes. I think what Skype announced with Motorola is representative of the fact that Motorola has Symbian interests as well as Windows interests. If the smartphone world is predominantly Symbian-based at the moment, one of the advantage of these things is that if you want to infect the system, you either have to get the mobile operator to say: “Oh - OK we will put that in.” Or you actually have to have an operating system where the user can decide to add Skype to the device. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA COOK Report: So if Skype runs on Symbian, then I as an owner of a Symbian OS smartphone, can put Skype on my smartphone regardless of when the cell phone manufacturer does or not? Henshall: That’s correct. I have software on Symbian-based Nokia smartphone that didn’t come with the product to begin with and just the other day downloaded the Opera web browser. I also downloaded a program called Agile Messenger. You are an Apple user so you won’t use this but there are some Apple equivalents for the use of multiple IM systems within one application. Trillium allows me to combine my AIM buddies, Yahoo buddies and MSN buddies on my PC. Agile Messenger on my cell phone, because I have an Internet connection on it, allows me to run ICQ, AIM, MSN and so on and to run them all at the same time. I keep wondering when they will add Skype to this mix. Here is their URL - delivered to you as we speak via Skype IM. http://www.agilemobile.com/agile_ messenger.html COOK Report: What I think I hear you saying is that with Skype being potentially Symbian compatible is that no matter whether Nokia or Motorola or Samsung, or any of a bunch of others, preloads Skype or not, the end user with a Symbian OS using smartphone just grabs Skype and sticks it on his device. By making Skype Symbian compatible, any user with a Sympbian OS smartphone who wants to do so can add Skype to his cell phone or PDA? Henshall: Yes, but I see two issues here. Anyone with a Windows compatible PDA can put Skype on that with the result that if you have a Wi-Fi-enabled PDA then for certain areas and certain sorts of events your PDA has become a potential mobile phone replacement. It depends on whether I have access to Wi-Fi enabled hotspot or not. The I-Mates that Skype is now coming preloaded on are PDAs that also have a GSM card in them as well as being able to have 3G/UMTS. [Editor’s Note: According to http://www. umts-forum.org/servlet/dycon/ztumts/ umts/Live/en/umts/What+is+UMTS_ index “Universal Mobile Telecommu- nications System”, UMTS represents an evolution in terms of capacity, data speeds and new service capabilities from second generation mobile networks. Today, more than 60 3G/UMTS networks using WCDMA technology are operating commercially in 25 countries, supported by a choice of over 100 terminal designs from Asian, European and US manufacturers. [snip] 3G/UMTS [is seen] as a key enabler for true “mobile broadband.”] Henshall: What Skype has done here is to open the door for the PDA to become a cellular replacement at very very low cost. Anyone who can offer you a Wi-Fi enabled smartphone can offer you very cheap VoIP dial out rates as part of a mobile device. Such a device is certainly a cellular replacement potentially for use in the home and certainly for use in hotels and airports. Now aside from such hotspot-enabled action, the place where mobile has the biggest benefit is when I jump in a car. Mobile works when I am driving around while I have to be stationary for Wi-Fi. Consequently there is a sort of tension that is emerging and symbolized by the Motorola deal announced this week and the I-mate deal of the week before. Motorola is going to provide a variety of hardware to a variety of companies but the consumer is going to sit there and say what do I really want in my handset? Really what they want is just one handset and one number. They want one place of focus or identifier. They may want to use it in multiple places as in I’ll take this call on my lap top now. But later on I might want to shift it to my mobile. I may even want my camera to ring. Changing the Way People Work Henshall: Skype, as you are beginning to see, changes the way people work. Way back in 2003, I was beginning to compare Skype to Microsoft Office where the office platform was all about text and email and things like that. Now I have begun to run off of two screens. A work screen directly in front and one with Skype as 14 an application off to one side. At the moment I have Agile Messenger on my work screen. On my left I have my Skype and email applications. This is basically a communications screen. This modifies the way one begins to work and helped me to see Skype early on as something that was creating a new communications platform in a way similar to that in which Windows created a work platform. But the things that Skype is doing all have the potential to pull users away from the old world of text and email to a newer world or real time chat, IM, voice calls - in other words integrated applications that are ultimately tied around your phone and hand set. COOK Report: Indeed, and tossing around URLs in the midst of an interview is an interesting experience. I note the other one you sent where Peter Cochrane the former CTO of British Telecom say that full adoption of Skype on a mobile smartphone sent his mobile bill down from $500 in one month to $10! http:// comment.silicon.com/0,39024711,39127 916,00.htm Now that as a signpost for the future would seem ominous for the phone company. Henshall: It will certainly change the phone company’s future. Ultimately it may change the way that things are charged. What Peter Cochrane’s report says to me is that he actually uses the phone more than ever. His call costs went hugely down but I bet you he is talking twice as much. What I want to ask is how do the hours in his day change? I play around with numbers a lot and remember watching last week as Skype went through 2,000,000 active users on line at one time for the first time and cranked out 3.5 million minutes an hour of talk time. We are looking at calling patterns here that are radically different from the traditional phone call. COOK Report: From my own experience part of the difference is that you are in front of your machine with all its tools and data and using a softphone program that is keyed to sending text and files to the person you are talking with during your conversation. It encourages you to browse the web as you speak and grab The COOK Report on Internet May June 2005 data, URLs or whatever and send it to the person on the other end whereas the Skype software captures the traffic for his review at the end of the conversation. It begins to take on aspects of a face-toface meeting especially since the sound is so good that nuances of voice traverse the connection very well. or not in terms of presence. Therefore I really don’t care whether there is a number behind that or not. Perhaps Skype should know that, if you are not on line, it should dial your mobile number. What does matter to me is that, when I want to use a VoIP line that costs less than cellular, some entity must provide number or name for the connection device. Henshall: Quite true. COOK Report: I note also you’re sending me http://www.oreillynet.com/pub/ wlg/6491 where the author is comparing Skype to a baby Microsoft; something that David Reed was hypothesizing about in early December. Henshall: Consider baby Microsoft as a hypothesis to weigh against what we are now observing. And then consider that in five years time this discussion about mobile may be completely different because the mobile may be able to project a keyboard on to your desk when you need it and it may be able to shine a 12 inch screen onto your wall in a projection TV mode of operation when that is convenient. Furthermore you may almost never put it to your ear because a tiny Bluetooth headset in your ear will handle the audio. Things keep on getting smaller - that much seems predictable even if form factors are not. COOK Report: If a Motorola wants to sell its cell phones, it seems that having Skype aboard them might be an inducement for the consumer to look for and choose a Motorola phone - although the mobile service providers can’t be too happy about that. True? Henshall: Yes. I think so. But the problem that I see as the greatest challenge at the moment is the issue surrounding my phone number. Ultimately I think phone numbers will be irrelevant. When I want to talk to you, I want to be able to click on your name. This actually already reflects what I do with my cell phone. I hit my contact list and scroll for the person I want and key in the first digit of their name and push call. Skype makes this even easier because it tells me whether you are even online Whether it’s in the SIP world, like Telio or Lingo, or whether it is in the Skypebound world, an important question is what number is it going to be? Skype can give me inbound numbers that are just like Liebertell. Or SkypeIn could actually be my mobile number. If I could have a working SkypeIn number that was my mobile number, I would pay my mobile operator an extra ten dollars a month for that privilege. It is a done deal because that is a VoIP access line charge. What happens is the same thing that happens when I put Skype on multiple machines at that point. My computer rings if it is on and if my cell phone is on, that also rings. If I have a choice, I answer the computer and take the call on VoIP. If I am actually home in my house and I have one of these combined I-mate GSM cell phones, my cell phone will know to take that call on the Wi-Fi connection and treat it as a VoIP call. COOK Report: Therefore the flexibility, independence and power of the platforms that are in the consumer’s hands are setting up a series of relationships where the patterns of connectivity are going to happen in ways that work economically and physically for the convenience of the consumer and not the cell phone company? Henshall: Let’s take the following as a thought experiment. Let’s say I need a new cell phone right now. I go down to Verizon or T-Mobile to have a look around and they say to me: “Stuart you can have this cell phone. If you want to buy it outright it will cost you $200 or $250. But it’s a nice top of the line version. But it doesn’t have all the bells and whistles - yet. But if you sign a one or two year contract with us we will sell 15 you that cell phone for $150.” And I say: “OK, fine,” and take a contract. But now we are two years down the road and I go out on the same mission. Only now I see all these fancy new Wi-Fi combo dual phones - wha ever you want to call them - out there. But they are all rather expensive. Like $500. That’s really expensive. I’d like the new features but I go back to T-mobile where I find out that I can get a contract for another two years and that the equivalent of the $250 phone only costs me $50. But if I get that new phone I can drop my land lane and that land line is costing me $300 a year. So I say: “OK, I’ll drop the land line and buy the more expensive phone.” But then I notice that I can get the same nice expensive phone from Amazon for $30 a month but more than that they will give you your old mobile number on the new phone and help you get everything transferred to a phone that can also do VoIP under the right conditions. With Phones Turning into Software User Convenience May Have Far Reaching Impact COOK Report: So some of the new business models out there are for those people who understand the complexity and can engineer flexible programs that will deliver value out of that complexity for the poor befuddled consumer? Henshall: I could say that I think there are two sorts of issues. One is the need for people to move to multiple phone numbers that was the by product of the creation of mobile phone networks reached by their own separate phone numbers. We now have Skype doing the same kind of thing. But we are now at the point with the technology where we could say - could we wrap this all into one again? If we wrap it all into one, we may get to the point where what is spent on the one roughly matches the sum total of what we spent on the parts before. For example I am happy to pay my mobile operator a little bit more if I can get rid COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA of my land line, fine. One of the things that Zennstrom said early on was that we used to think that the ability to have a conversation was tied to a piece of hardware involving wires. This is no longer true. Phones are turning into software. COOK Report: As Rich Shockey said: “Voice is just a bucket of bits.” Henshall: Right. And the question becomes where does the bit bucket or software reside? In what type of device? This type of software is just going to infect every device in some way. The key gift this next Christmas potentially is a Wi-Fi capable phone handset that looks juts like your normal phone handset except that it is Wi-Fi capable and you have a VoIP client on the other end of it and when you want down to a local hotspot or go next door to the neighbor’s, you just take your phone with you. This is really just a low cost version of the HP PDA. It says how do I take an HP Wi-Fi enabled PDA and engineer the cost down so that it costs me $120 and how do I connect that to some VoIP provider? Do that and you have a very low cost handset that you can take anywhere there is Wi-Fi. COOK Report: Well in this context, can we review the basic mobile phone protocols? Henshall: ATT, Cingular and T-Mobile are all GSM. Sprint is PCS, and Verizon is CDMA. Verizon has also introduced into certain high density markets EVDO which is their version of 3G. They are charging $80 a month for EVDO - a rate that I certainly don’t find attractive. If I have to make a choice that when I need a bit of bandwidth, I can get Wi-Fi for nothing or do I want to pay $80 a month for always-on connectivity? For certain people that may be a good deal, but for the majority, I find it hard to justify. COOK Report: But all these protocols are good enough to get the basic tasks for which people use smart phones, voice, email, web and so on done? Henshall: Yes. There is not much difference. Certainly we have seen that minutes have been migrating to mobile for quite a long while. As rates come down, people are using their mobile phones more and more to talk. Of course, use varies by area and calling scheme. In America I think we are relatively unusual in that that when someone calls our mobile phone that in-bound call costs nothing extra. In Europe however ringing someone on their mobile is a very expensive long distance call. The same thing in New Zealand. You can get an 800 number on your mobile and make it cheap and easy for someone to call you, but doing so is a big barrier for moving from one system to the other. Typically within those markets your monthly plan gives you ‘x’ minutes of calling time within the network. They operate under a very different charging structure and this is why SMS and text messaging are so big in Europe. There it costs only a few cents to send a text message as opposed to a considerably more expensive voice call. COOK Report: This reminds me of the note on your blog of about ten days ago wondering if Skype would enable free SMS messaging? Anything new going on there? Henshall: Let’s just say that I think that the interesting thing is that, once you can go SMS to mobile or from Skype to SMS and go from Skype to mobile with a text message to find out if someone is available to talk and can receive a response, it means that you can call that mobile effectively at Skype-Out rates which in some cases may be very cheap. COOK Report: Because Skype is so good as a bridging technology, do we have something going on here where Skype can go up to a manufacturer and say let’s do a deal where we preload Skype on to your mobile phone, as long as doing so costs the device maker virtually nothing. The device maker is motivated to comply because having Skype there will make the device more desirable to its customers. But, on the other the mobile phone company that has to offer the Skype loaded handset to its customer must be not nearly so happy? Yes? 16 Henshall: I think the traditional answer is yes. But I think that logic needs to be overturned along with the belief that products like Skype are bad for the mobile operator. The mobile operator after all is about mobility. The question here is always-on. Always ready to talk. 3G is about always-on and so is Skype . You could say that Skype is just accelerating that style of working. I can sit here in my office and connect with five other people in a conference call. We can all put ourselves on mute and leave that call running the whole day if we want. Anytime we want someone’s attention we can just un-mute our headset and call out to them. They call it “push-to-talk.” I can have that up right now on my computer namely a “push-to-talk” network for up to five people and continue to run another iteration of Skype, side-by-side with it, and continue to take and make calls there whenever I want. Skype as a Digital Convergence Tool Box Dependent on Broadband COOK Report: Skype, in the sense you describe it, is beginning to look like a Garden of Eden for someone who likes to play with a kaleidoscopic range of possibilities of the technology tools enabled by so called digital convergence. I think what you are saying is that for a seriously new piece of software like Skype, once you follow its emergence and apparent strategy, assuming you know business use of communications very well, the number of possible ways that these tools can be combined with other tools to open up new ways of doing business becomes really impressive. True? Henshall: Yes. Definitely. I am looking for two other links as we are talking. One explains a Nokia launch of a voice messaging type product. The article explains how Nokia is enabling the capturing of a voice message that could be edited and sent to a podcast or an audio blog. http:// digital-lifestyles.info/display_page.asp?s The COOK Report on Internet May June 2005 ection=platforms&search=1&id=1916 [Editor’s Note: The article at the above URL explains: “When we initially learnt about this, there was a degree of ‘Oh, fancy voice mail, OK,’ but looking a bit deeper, we see this could be a significant development for podcasters and audio bloggers. Rather than just phoning in your comments directly on a phone line, the ability to trim and edit the pieces before postings them means they can be polished before sending. “It’s well known that Nokia is to include their LifeBlog software with their new 7710 handset. Xpress audio messaging could become another tool in the podcaster’s arsenal, enabling podcasting on the move, without a PC. It will all be down to the power of the audio editing software. We’re keen to get our hands on it to see if our hunch is confirmed. “For the straight messaging, Nokia clearly hopes that this will give them some leverage in the highly competitive - and lucrative - youth mobile phone market: ‘Nokia Xpress audio messaging enables operators to differentiate their service offering from competitors, by utilizing existing infrastructure,’ explains Juha Pinomaa, Vice President, Mobile Phones, Nokia. ‘For consumers, Nokia Xpress audio messaging combines ease of use, affordability, and adds a personal touch to greetings, congratulations, or allows to share a special moment like a grandchild’s first words.’ “Recorded audio messages can be sent to all MMS-enabled GSM handsets and stored and replayed as easily as any other multimedia file, and Nokia will be introducing support for legacy phones within its MMS solution.”] COOK Report: I am seeing that this really is something that you have to do before you can appreciate it fully. When I went to broadband two years ago, it was to get Vonage and save money on my phone bill. But I am now beginning to see that a new application like Skype is making broadband into something of a platform that can serve as a foundation and a communications hub for many new applications that enable collaboration of one sort or another. If I were selling broadband, I would look at this information and find that it would warm my heart because it certainly broadens the market for my product. If you have a decent phone line, you can do a lot of things including most of the web with dial-up. But for these sorts of things you need broadband. Yes? Henshall: Broadband is the foundation technology for all this and you are right in that it certainly is the underpinning that makes all this possible. Proprietary Skype, Open SIP and the Enterprise COOK Report: We have done a pretty good tour of the consumer small business world for these tools, but it seems to me that there is likely another huge world inside the enterprise. However this is one where security concerns will indeed run paramount. The question in my mind is whether this enterprise universe has to be thought of in a very different way. Is what we have occurring two separate streams of development? One stream inside the enterprise and one for the rest of the world - until Skype begins to infiltrate the enterprise? Another interesting aspect is that with Layer One and Two user-controlled optical networks, you need a fair sized corporation to implement this technology with this. It’s you and I, on the outside edges who are way ahead of many of the enterprises. ones, and think OK how many of your employees are already allowed to bring a mobile phone to work? How many are using AIM on the corporate network? How many are allowed to carry a USB stick into and out of the business? I think that for enterprises there are two separate questions. One is how do they establish operational procedures governing trust of their employees in general? The other is that we know that the majority of companies are tracking emails and who sent what to whom and are able to look in on almost everything the employees do while Skype at the moment makes it more difficult for a company to see who is on an employee’s buddy list. The information is available. COOK Report: This is because the Skype API is now available? Henshall; Yes. You can use the API to generate a list of calls made, the time of the call, the person who was called - all that. You could use the API to add a plugin to Skype that said as long as these two things are registered on our corporate network and working together, the Skype client will be reporting to us at all times what this person does with it. Whether or not they are sharing files, or whom they are chatting with. That is one way to get a view and I think that Skype will probably generate some enterprise systems that are capable of doing that. If not, it is now a simple programming exercise. It seems that Skype could be quite advantageous inside the enterprise, but there are some security issues that will make Skype’s entrance there into a difficult proposition. What can we look forward to here? There are some things in Skype that have never been done before. I registered my name as Stuart_Henshall but if I had wanted to register stuart@henshall.com, I could not do that. At the moment, the @ is a disallowed character. So are some others. Now as far as I know with every one using SIP, they capture your name but also your email address. Presumably a premium Skype product in the future could allow you to use your email address as your Skype name. Henshall: There are some security angles from the enterprise point of view that I am sure I don’t adequately understand. Skype is obviously creating a debate on those issues. I also look at enterprises, especially the untold number of smaller COOK Report: Let me go back to what David Reed said early in December and that was that Skype was doing all the right moves in comparison to SIP which just wasn’t getting much of anything done. Nevertheless Skype being propri- 17 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA etary made him a bit nervous. At the same time he said that he felt their business strategy moves were impeccable. He predicted that this combination could turn Skype into the WinTel monopoly platform of voice communications. I am wondering two and a half months later as things continue to unfold in a quickening pace how you view that tentative surmise? Henshall: I think all of us, if we had a choice, would like to see a completely open source Skype competitor. But let’s just flip the question around and say that they have indeed made the right moves. They have launched an open API that relates to text and all other sort of messaging capabilities. Why don’t we ask and request from the phone companies that they share their APIs and open up their networks in the same way so that what we see is an innovation market - an ideas pool that is created at the juxtaposition between the convergence of mobile, landline and Skype and let the real innovators get in that field and really create the products that we actually want? COOK Report: That is a very provocative chain of reasoning and I think the netheads would hear it and snort “impossible because after all they are phone companies and phone companies will never do that.” But on the other hand you have to wonder whether Siedenberg, if he has any sense, may realize 6 months from now or a year from now that if his ILEC is going to survive, he will have to start thinking like this. Is that part of what I am hearing? Henshall: Yes. I think so. I don’t know all the depths of the technology that exists within a mobile operator. But let’s use one example. The mobile operator has great location information. Fine. Give me that information on the API and let me now dock that with Skype profiles. Let me work with advertisers or other things and people that start to make sense. COOK Report: It sounds like Google has been doing some of this? Henshall: Google has been very smart. But in fact I think Amazon has been even smarter with their API. In a world where telecom more and more is just software and where you are an older telecom company trying to compete, all I can say is where are your APIs? And seeing that you can’t think through the solutions yourself, why don’t you turn them over to the open market? COOK Report: As we think every now and then about where all the jobs are going, I am beginning to form an opinion that the place where a good deal of business opportunities are to be found is in putting the pieces of this kaleidoscopic puzzle together in new and unexpected ways. Right? Henshall: Well, if your were SBC and you had just bought AT&T and you were sitting there saying now we have acquired all these engineers and we will now have to get rid of them. Why not give them something useful to do and turn them loose on the APIs? Skype versus Peerio - Architecture Implications COOK Report: Thinking about P2P IP telephony and its possible evolution, let me ask about Dmitri Goroshevsky, Peerio, and a possible federated architecture for Skype or a Skype clone. What Goroshevsky was telling Martin Geddes in his recent Telepocalypse blog was pretty fascinating, but, on the other hand, if Richard Stastney’s violent reaction to it on my mail list is accurate, Goroshevsky is full of hot air. What do you think is going on? Henshall: It’s a good story. I have heard it, just like Richard Stastney, making the rounds for months. So far they have shown NOTHING. (Note they did announce at Internet Telephony. However, I still can’t download their app. Nevertheless, these comments could become very old very fast.) COOK Report: The architecture that he describes could be done but until someone does it, forget about it? 18 Henshall: Right. I don’t know whether it is smoke and mirrors to try to enable them to make money or if it is the real deal and they actually have the chip. When I was at VON in November 2004, they claimed they had the chip and were taking it into production and their device would be available by now. But they have said this too many times and then failed to deliver to have much credibility left. However let me also say that I think that it is quite plausible that the structure for IP telephony that Gorsohevsky is talking about could become the real future for communications. COOK Report: So one might well say that the message here is that while Skype so far is very successful, it certainly has some weaknesses. That from the work of Henning Schulzrinne’s students and others we certainly have a good idea of how it operates. [See http://www. cs.columbia.edu/~library/TR-repository/ reports/reports-2004/cucs-039-04.pdf for an early attempt at a Skype protocol analysis.] Therefore, folk should not sit back and assume that Skype in its current instantiation is the be-all and end-all for peer-to-peer IP telephony? Right? Henshall: Let me answer by asking a question about what Goroshevsky is talking about. It seems that he is saying that he will give anyone who wants one an exchange and that with an exchange those people are free to go and connect up to an emerging mesh. Now here’s my question: What happens if Skype says that the only way to solve the enterprise’s problem is to give the enterprise that same capability and that you as an individual can also buy the same capability - so that rather than having one log on (authentication) server in Denmark, anyone can have a log on server. In this arrangement the authentication servers mesh with each other and then you make your decision as an individual as to whether you want to join all servers or whether you want to join just a few subnets. COOK Report: What you are also saying is that if the Skype people are as intel- The COOK Report on Internet May June 2005 ligent as we would assume them to be, they have to have resources and brain power devoted to work on these kinds of enhancements. Henshall: Sure. Every enterprise has to want to be able to share their presence information internally. They don’t want to expose everyone in the company to the world-wide network. They do want to expose people to their value network and sharing presence within that value network makes a lot of sense. They must confront an assumption that says: “Hey Skype’s got this wonderful peer-to-peer network, and everyone has to go through this authentication server in Denmark when they first log in just to make sure they get hooked up to a supernode.” One might well ask: “Who says that this has to be the long-term strategy?” Who says that the enterprise or even you and me cannot actually run one of those things (a Skype system) and do we really understand what happens when we do that? Because all Skype has been doing is to sit there in this new, so far imaginary situation, saying to an enterprise: OK you can now run your own Skype system and select the network of people who will be permitted to attach to it. We will also give you the ability to select from your network the people whom you want to expose to our network or to any other network. At the same time you can make your own decision about how your network will handle inbound calls from our other Skype networks and all those sorts of things. COOK Report: While Skype’s peer-topeer architecture gives network security engineers fits, it is this same architecture that would for the most part make it immune to network failures and would also make it difficult if not impossible for any centralized telco to shut it down. While the authentication server is a single point of failure, what these technologies have done nevertheless is close to the state-ofthe-art in being reasonably failure proof. Would you agree? Henshall: Yes. And the other component that follows for the enterprise then, at least ones that operate at multiple global locations, effectively looks a bit like a SIP server for a SIP system - if I have a laptop and want to log on to my home office. Here are my buddies I can see on line - all of which is fine, but I am connected through to them not via a log on to the global Skype, but rather via a log on to my company’s Skype. COOK Report: So are you saying that if I am Ford Motor Company and I want to set a Ford Enterprise Skype System that is absolutely independent of the public global Skype system, this is possible? That you could set it up on the enterprise net so that it ran its own authentication server and its own supernodes which were absolutely independent of the global Skype server and its nodes? Henshall: Well, why not? Really? You could also tune it to select whatever outsider parties you want to share presence with. Or you could say: “Stuart, we trust these particular sets of people with every thing, and so we want them to be able to log onto us but also be able to log on to the public Skype because they needs to share their presence globally.” COOK Report: And in a situation like this gateways, become really important. Henshall: Yes. Although potentially I assume that because I am in the enterprise I automatically assume that my call would be routed out through the authentication server. COOK Report: But someone in the enterprise will decide what the policy will be for that server and how it will be implemented? Henshall: Why not? And for large-scale enterprises you could set up an enterprise SkypeIn such a way that the administrator may turn certain features on and off. You don’t want file sharing to run on your Skype? OK, we will turn it off. COOK Report: But probably somewhere some person in the enterprise is going to say: “Doing this looks attractive but until we see the source code for all of this software we can’t afford to take the chance of using it because how can we be 19 absolutely sure there isn’t a Trojan buried somewhere inside the code?” Now is this something that a nuclear laboratory has to do? Is it something that Morgan Stanley has to do? Does Dunkin Donuts have to do it? Probably not? Henshall: They never did it for their Microsoft software. COOK Report: Right! <laughs> But maybe they have learned their lesson? You see in a general sense what I am saying. But you don’t seem to see it as a showstopper. Henshall: There are an awful lot of organizations yet to be VoIP-enabled in any way. They are smaller businesses with all these PCs that they constantly have to upgrade. In such a business Cisco comes along and tells it that it needs to spend $100,000 on an IP PBX and related Cisco VoIP phones. It is going to be met with incredulity. The voices will say, like for what? For example, are you telling me that my small import business that communicates with Asia all the time or my small travel business that needs to book bed and breakfasts around the world, are you telling me that with the PC’s I have and a high speed Internet link I can do all that voice communication cost for zero? Thank you very much. You’ve got a deal. Skype Marketing and Performance Research COOK Report: Focused on the issue of business opportunities created by Skype what are you able to share with me about your interest in creating a Skype user’s community - an interest that you mentioned in our first conversation a while back? Henshall: I am close to starting an online blog designed to focus the interest of the Skype user’ community but also to try to retain the independence and input of developers, counterpoints and things like that. At the moment I am still experimenting with it behind the scenes trying to understand what sources for different COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA sorts of information are out there. I think Skype really understands is one of the Skype strategy. That could introduce there are potentially many spin-offs that most valuable: It is all about presence. strategic foresight with stretch views of could be gained from bringing that com- While Skype offers free person-to-person where it may go and where you might munity of people together. calling, the larger question is what can go. Ultimately the value is in creating you do when you have presence? For new options. COOK Report: It sounded to me like example, the match making, the dating you were saying that if you could bring industry at the moment is worth well COOK Report: So the dialogue would go enough people together you could cre- over a half billion dollars a year. something like: if this is your business, ate an effective counterweight against and this is what you are doing, then these anyone who wanted to take Skype in a COOK Report: I remember doing an are the tools that we have now that can be direction that would be bad for its users interview with Henning Schulzrinne in worked with. Right? Do you agree? Yes September 2000 about SIP and presence or no? And if this is tool set, then what as a whole? and Instant Messaging and their impor- are the variables in the things that you Henshall: It’s a nice idea! What I have tance and here we are almost 4 and a half need to be thinking about so that you can in mind is a bit like Consumer Reports years later and we still don’t have those get underway? or JD Powers. problems solved - or so it would seem. Henshall: As a part of this, I have been Unfortunately Skype is really not a mar- Henshall: On the issue of SIP, while I working in the background trying to figketing organization at all. I encouraged am not a SIP expert, my understanding ure out how to set up research programs. them to start their forums early on. Be- is that if you want to use SIP and add an I certainly believe that whether you are cause of those forums they have a strong Instant Messaging chat system to SIP- Motorola or Nokia or whomever they core group of beta testers out there. These based VoIP software you have to use all need to buy more research on Skype people have become absolute maniacs in something called Simple. Simple is the users. They all need to know more about helping Skype develop its product. Some standard for building chat into SIP. I am Skype and the cheapest way to do this is of them are retired, some are young kids. not certain how well this works. Howev- to buy part of a global research package. The testers are a crosssection of people er I had a very interesting experience the Questions like how many Skype users who have a passionate interest in what it other day with this new company Teleo. last month dropped landlines are of obvious interest. However it is deeper prefercom that just launched. is enabling. ence information that really matters. COOK Report: Do they have a commu- It looks like a Skype clone, but it is built nity of 3rd party developers like the iPod on SIP. The first interesting thing that COOK Report: Do you envision using has acquired? Would part of your idea be I observed is that when you search for your blog to generate tools that could to identify, coordinate and interconnect someone on the network and add him as help collect data and that you would some of these entities where your role a buddy, it doesn’t ask you for authoriza- tabulate and report on, in the manner becomes one of helping the developers tion. Therefore on this system, if you are of a Gartner group, for Skype reporting create viable business models for use of on the Telio network and I search for agency? the Skype technology? You are the guru you, find you and add you to my call list, who knows how to build what will work I automatically now know at any time Henshall: The blog won’t generate the tools. This is a discussion that needs to in that world? Or I would come to some- whether you are online or offline. one like yourself if I am trying to figure be had with potential customers. out how to best apply this technology in Imagine if this system became popular a way that supports my on-going current and you were a telemarketer, you would COOK Report: But, until they fully unbusiness? just add the whole system to your direc- derstand what is going on, they may not tory and boom you have all the users at even realize that such research is imporHenshall: Yep. All this would be key your mercy. tant to them? elements of what can be done. I believe that we are still in the very early days of COOK Report: Is part of what you have Henshall: Even formulating the queswhat the Skype API is and where people in mind a role where in the future a com- tions to ask may be useful. For example actually are. But I think things could ac- pany like this might pay you to advise once someone becomes a Skype user celerate very quickly - especially once them why making their software respond (colloquially we call them “Skypers”) the first real application that solves pres- in this way is not a good idea? how long before that person begins to ence comes out. This will provide a wake think about dropping a landline? But Henshall: I am committed to being more there are other issues as well, clustered up call to more than a few people. targeted in providing seminars or events around the question of what sort of prodCOOK Report: How do we know what in this area. I’ve run both Scenario and uct would you like? Is a landline USB to look for? Innovation programs for the largest com- handset that is semi Skype-enabled but panies. One way would be for a company otherwise looks like an old phone what Henshall: The part that I don’t think to test their strategy against my perceived you really want? Such a product certainly 20 The COOK Report on Internet May June 2005 lacks imagination. COOK Report: With Apple’s interest in good design I have to wonder where the Apple I-Phone is? Imagine if this system became popular and you were a telemarketer, you would just add the whole system to your directory and boom you have all the users at your mercy. Henshall: Indeed, and if Apple doesn’t do it, you would think that someone would make a little GSM phone-type gizmo to plug into the top of your iPod and lo, it’s a phone. It has the power pack in there. It has the data. Why not? Schulzrinne: If true, this seems like a really bad implementation, which other implementations have fortunately not followed. Henshall: Thanks for you comments. Symposium List Exchange Between Stuart Henshall and Henning Schulzrinne Editor's Note: On March 1, 2005 Henning Sculzinne commented of the interview with Stuart Hensall. The exchange follows. Henshall: On the issue of SIP, while I am not a SIP expert, my understanding is that if you want to use SIP and add an Instant messaging chat system to SIP based VoIP software you have to use something called Simple. Simple is the standard for building chat into SIP. I am not certain how well this works. Schulzrinne: Well, he could start by downloading Microsoft Messenger, which has supported (pre-standard) versions of SIMPLE for about two years and is pretty reasonably standards-compliant now. There are open-source clients as well (GAIM), not to mention that HotSIP had a client at least three years ago. Henshall: However I had a very interesting experience the other day with this new company Teleo.com that just launched. It looks like a Skype clone, but it is built on SIP. The first interesting thing that I observed is that when you search for someone on the network and add him as a buddy, it doesn’t ask you for authorization. Therefore on this system, if you are on the Telio network and I search for you, find you and add you to my call list, I automatically now know at any time whether you are on line or offline. I’m aware that MSN is “close” to using the SIP SIMPLE standard. In the MSN environment which looks as closed to me as a consumer as Skype it sort of works. Sort of because it fails to connect as well as Skype in my experience. Schulzrinne: There is a common confusion between MSN and Microsoft Messenger. The latter works in any SIMPLE (and SIP) environment, not just MSN. (MSN had a proprietary protocol, with a transition to SIP/SIMPLE scheduled.) Henshall: From a user perspective it’s IM and voice and video etc. For the most part IM works, voice even if it connects remains “poor” quality. I can’t rave about it as a product, and I don’t have any buddies left that want to use it in preference to Skype. Schulzrinne: We find it quite useful for application sharing and for integration into a bigger environment that we can control (in terms of reachability and features). That may not matter to you. Henshall: I’ve tried to use Gaim on three different occasions with each upgrade and it has failed in each case. I’d add the user interface and ease to sign up / on are additional failings. Frankly it’s just clumsy and ugly. Is this a result of SIP needs? Schulzrinne: No, this is a result of an open source project done by volunteers that tries to integrate multiple presence systems into one client. Henshall: This again is not necessarily a reflection on SIP. However it never worked for me. We also have executions 21 of SIP clients like X-Ten; I think they are also clumsy and failing to step forward. I’m happy to put the best SIP client to the test in a consumer environment. We should also look 24-36 months out. The Teleo information quoted below was correct when I blogged it. Since then I’ve removed Teleo and have ended up with a couple system restores in the last week. Today I embraced and tested another SIP client. Just launched. http://www. damaka.com/ Damaka. Schulzrinne: It claims to be a SIP client, but our protocol analysis makes that appear dubious. Henshall: This ostensibly is another SkypeClone. At the moment in my view it is another failure. Currently it doesn’t offer any inbound or outbound number options it is merely a SIP client that uses GIPS voice engine again. Schulzrinne: It is interesting that all your preferences are for a vertically integrated service provider, i.e., somebody that provides software (and maybe hardware) and termination srvices. Have we come full circle, back to the AT&T days where you could choose any provider as long as it had a blue globe logo? Henshall: Unlike Teleo it does require authorization however that activity required rebooting after adding my buddy (his system crashed twice before we connected). Then the voice quality was Skype quality. It failed to offer more than a two line option, the interface is clumsy (compared to Skype). Both the last two items look to be at least a year behind Skype in development, no matter how much money is thrown at them. How can Damaka hope to even compete, Skype is gaining 150K downloads per day. I may be fusing the elements of design and user experience with SIP in my comments. However these recent releases suggest to me: That GIPS and SIP solutions must be closing on commodity ease to bolt them together for PoIP like solutions. That anyone not designing for best in class is simply kidding themselves and best in COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA class for the next round is not what we have today. It has to include 3D audio solutions. Want a neat sound experience? Try SmartMeeting out of Sweden. Next is Video, which I’m sure Skype is close to rolling out. Schulzrinne: Microsoft Messenger (among others, such as our own SIP client or eyebeam or the SIPquest client or Hotsip) already has video. Henshall: Then it also requires new thinking on Presence. We need more that online, away etc. Many neat experiments have been done. Schulzrinne: Indeed; and standardsdriven rich presence is likely to roll out shortly; see http://www.cs.columbia. edu/sip/draft/rpid/draft-ietf-simple-rpid05.html for one part of this. Henshall: SIP’s innovation time line appears to be tied to “hardware” platform thinking while the new / emerging solu- 22 tions are focused on “software”. IS there a development cycles gap? Schulzrinne: Since SIP hardware and software vendors are not vertically integrated, they go for the money - this is currently mostly enterprise systems, as adjuncts to IP phone systems. For reasons that I don’t understand, the guiding design principle seems to be skinning to make it look like a 1950s space-age-impression phone. Symposium Discussion December 1, 2004 to January 4 2005 Sype versus SIP: a Debate Between the "It Just Works" Point of View and the Standards Based, Interoperability World View Highlights Editor’s Introduction: I knew that I should get back into VoIP sooner or later but for quite a while I didn’t see anything to peg the story on. However, thanks to the Skype thread that Dave Hughes started on December 1, 2004 and to David Reed’s comparative analysis of Skype and SIP, I think we have the elements that will help to bring some clarity to a wider audience. Dave Hughes Discovers Skype On December 1 Dave Hughes: I don’t whether anyone is interested on this high tech COOK Report mail list in a humble VoIP test that works better than it should be expected to, given its zero cost. Two days ago I shook down my (10Mb) Skype software on a Windows 2000 laptop with a very good quality voice conversation with Gordon Cook and his Mac based Skype. Tweaking my sound, with and without built in microphone and speaker to get the right combination for me. I knew the sound with broadband at both ends here in the US should be good, as advertised. It was. But so is Vonage (Lingo less so), and other IP point-topoint VoIP phones. But I wanted to give it a real test from here in Colorado to far off Namche, Nepal where it would have to transit, not only US broadband, but one satellite IP hop from Hawaii to Kathmandu, and then a second satellite hop from Worldlink, Kathmandu, to Tsering Sherpa’s Namche, Nepal “Cybercafe” - with a minimum latency of 1,200ms. With the added problem that during the off-climbing (Mt. Everest) season, he can only afford 64kbps (about $800 a month) feed to Kathmandu. So typical VoIP breaks down with any load at all, all the way to Namche. I am here to report Skype, with its rated 33kbps (which they also brag will work over a PPP dial up connection at least with 36kbps speed) worked very, very well. Much better than I expected. Very good voice quality and volume, (at least as good as a PSTN US call!) maybe two transient 2-3 second burps in over a one hour talk. He was on a laptop with plugin mike and speaker - me with mine. Full hour plus talk, PLUS a concurrent Instant Message exchange while we were talking (no effect on the voice). I didn’t even really notice the second and a half latency though it was there. Super! And NOW we will extend the link 5 more miles through the Linux server at the base of the Satellite Namche Cybercafe) location (11,500 feet), with three outdoor WiFi relays hops (one via a 13,000 foot high Buddhist Monastery) to Thame, Nepal, a D’Link switch, feeding two PCs where 9 Sherpa kids are trying to learn English from Mingma Sherpa in Pittsburgh via his DSL connection, BOTH email and Voice (he of course handles Sherpa, Nepalese, and English well, being 9 years a programmer in the US and grew up there in Namche). Because the bandwidth was constrained off-season, a standard VoIP IP Phone instrument connection broke down. Just takes too many bits with G7291, or G711ulaw or G711alaw codecs - purportedly from 30kbps to about 60 for higher quality voice. (Only Lingo will, under special circumstances install a G723 type codec that is supposed to go down to 20kbps - but given the questionable quality of voice even across the US with broadband, I’ll pass, thank you). I think Skype beats them all in voice quality/low bandwidth. And also with its ability for you to make a Euro credit card deposit from 10 ($12) for 10 hours of Skype to PSTN calls anywhere in the US, Canada, Western Europe or Australia for 1.7 euro or $.02 flat rate - only 23 charged per call. Gordon reports excellent quality at that charge rate to a call he made to Europe last week. I gave Tsering Sherpa who is hardly a high techie on the slopes of Everest simple email instruction to download and install the Skype. Says he got it done in one hour - most of that being download time. I am as impressed with the easy to understand software interfaces as with the voice capability. My Skype icon (on right now) reports 1,317,028 users online right now (meaning the program on their computer is awake, ready for a call) Reaches over 2 million during the day, drops to 800,000 or so at night. About 40 million free downloads have been made of the software. [Editor: on February 11, 2005 Skype reports 1,782,617 users on line at 2pm Eastern Standard Time. It is claiming 140,000 downloads per day.] So when he said: “”I knew it was over when I downloaded Skype,” Michael Powell, chairman, Federal Communications Commission, explained. “When the inventors of KaZaA are distributing for free a little program that you can use to talk to anybody else, and the quality is fantastic, and it’s free - it’s over. The world will change now inevitably.” This confirms my view that probably the greatest Internet Killer App for the other 5.9 billion people on this planet, ALL of whom can talk voice over a phone, even if they are illiterate for email or otherwise, is going to be free, cheap, very simple $5 SIP voice devices (not a complete computer required), connected unlicensed wirelessly. Eat your heart out Verizon/Cingular. And even Vonage or Lingo, much less Qwest offerings of VoIP - a billion dollars short and a decade late. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Ron Yokubaitis: One of my sons is an Exchange Student finishing his last year in Electrical Engineering in Belo Horizonte, Brazil. We talk weekly via Skype with very good quality. At least equals my AT&T cell phone for quality. Same $.02/minute. Jack has a low speed ( > 384 Kbps) DSL circuit in Belo Horizonte. Jack will text message me via cell to see if I am online or to schedule a chat. Text message is $.10/ message How can I get my T-Mobile “SideKick” (HipHop) to do VoIP? It does mail and web well. Instant messenger feature has be scripted by our software development group to run on IRC chat networks. Also our guys have a secure shell up so I can login to our Unix servers. Skype’s $.02/min. would sure beat T-Mobile’s $.30/min. voice rates. December 2 Damien Wetzel: Hi, Dave, I totally agree with you. I remember having downloaded Skype first, just one year ago, they were at 1 million downloads then. I thought what is Cisco-like VoIP worth, when you have this kind of software? Now if Wi-Fi networks develop quickly in a near future + Wi-Fi enabled smartphone. I think that mobile operators should be very worried. December 12 Dave Hughes: Hurrah! I just KNEW that Skype would work if nothing else would! I just got a Skype call, from Tsering who was all the way over at 13,000 foot in Thame, Nepal, (12 hours difference - Sunday night here, Monday morning there) in the classroom with 9 Sherpa kids. Five miles and three Smartbridge Wi-Fi radio relay hops from his Cyber Café base in Namche (on the Everest trekking trail) then over TWO satellite jumps to the US, (one with only 64kbps bandwidth) then over the net in the US to me! VERY clear! Then he called Mingma Sherpa in Pittsburgh from Thame, had a good clear talk. Mingma will now be able to teach those kids English, ORALLY, and not just by email, as well as use the link to instruct them in spoken Nepalese, how to better operate their classroom computer which Jim Forster of Cisco donated last year! Hey those kids are gonna get educated in spite of their remoteness! Because Tsering was using the built in microphone in Thame, I could just hear feedback from what I was saying. It seemed a long delay from the time I said something until I heard it back. You bet. It takes SEVEN seconds from the time I speak until it goes there, comes back (over 88,000 sat miles) and I hear my own voice back here! Why SIP Has Failed David Reed: Well, I have gone on record (at the MIT CFP working group) as saying that SIP may have missed its window, because of Skype. (And I was a big fan of SIP’s potential). SIP could have been what Skype is becoming, but the SIP community has been trying to replicate the walled garden before deploying. They are destroying the value of open interoperability that was in SIP, just as Skype is opening its APIs to get the boost of third party developers. SIP should have won, because it is an open standard, but the desire to create a business model that captures the old unsustainable voice revenues of the RBOCs has seduced Cisco and its customers into waiting and making the standard more complex. Unlike the old days of the Internet, where interoperability was the centerpiece, the likelihood that a SIP phone will work with one from another vendor is near zero. There was a reason that the major IP trade show in the early days was called “Interop”! So now instead of innovating to make SIP work as simply as Skype does out of the box, the business strategy of the access providers is: attack their best customers by finding reasons to block Skype traffic. This reminds me of the IT departments who tried to keep department managers from buying Apple II’s because they 24 were afraid that their budgets and power were at risk. Also reminds me the suicidal behavior of the Bluetooth consortium - (in contrast with the 802.11 vendors). Sending voice over IP is trivial. That’s not the technical problem. Getting scaled adoption is hard, and a common standard that works simply was required. SIP could have been a contender. It isn’t going be. And I think its own “proponents” killed it. This is completely analogous to what happened with Unix vs. Windows. (It is balkanization vs. a common platform). Open platforms can win, but a group’s self-interest in cooperation and coordination is often poorly understood by the members of the group itself. Linux, on the other hand, seems to be growing (Linux is now much larger than Macintosh, in terms of desktop market share. In server market share it’s been dominant for longer). What makes Linux win is that the group’s interest in interoperability is identified and managed, as opposed to ignored and frustrated by its own members. Steve Heap: Aren’t we comparing two different beasts here? Skype is an excellent system for making voice calls and IM relationships from one computer to another. Nothing wrong with that. The rest of telephony is about making voice calls from telephones to telephones/computers, and has a massive numbering structure in place to allow that to work domestically and globally. Skype have nicely bypassed all those issues of numbering by making up a scheme of their own, but that pretty much means that you cannot call a Skype customer from your home phone, payphone, cell phone, etc. It is hard to see how you will ever be able to do that unless they can persuade the world to introduce a new Skype international code. SIP is a solid protocol addressing the telephony (and other media things) issues. It is being introduced in large-scale installations. It carries VoIP calls within carrier networks, and increasingly between carrier/service provider networks. I don’t understand how you then can say that it has missed the boat as a protocol? The COOK Report on Internet May June 2005 people to that on the other end. Reed: SIP is “Session Initiation Protocol” - it’s not the protocol that carries the voice. It’s the call setup protocol, focused on user-user negotiation. And that’s what Skype does, at its core. As far as SIP dealing with legacy numbering plans, etc. since when does telephony correspond to 10-digit numbers assigned to instruments with microphones and speakers? That’s like saying driving a car is fundamentally dependent on driver’s license numbers. Numbering is an epiphenomenon to telephony, and in any case, SIP itself doesn’t define numbering plans at all. There’s confusion of SIP with the business model of (say) Vonage, which focuses on selling numbers and directories and gateways, since they don’t sell bit transport (you buy that from Comcast or RCN or Verizon DSL), and they don’t sell codec capacity (that comes with your PC or the ATA-186). 99.9% of Vonage calls do not involve two SIP endpoints. Vonage doesn’t even optimize for SIP-SIP calls, nor does it offer SIPs capability for additional services (like IM, presence, etc.) Will it ever? So it may well be that SIP is used internal to traditional phone networks in a transparent, trivial, or degenerate way. But that’s like saying ISDN was deployed in the US, when the only use it was put to was 128 kb/sec. raw bit pipes for Internet access. None of the AIN (intelligent network services) were ever successful in the US. Yeah, ISDN got used. But it failed. What also comes to mind is the call center business. I remember a company that hired people that were agoraphobic, handicapped, etc. for Avis/United Airlines call center work. If there were a gateway to Skype to reach voice end lines, it could save those call center company people a lot of money. It is a real solution if available for an industry that could use that. Heap (to David Reed): I agree with everything you say below - but then I don’t understand what you mean when you said that Skype has pretty much killed SIP? Do you mean in an end-to-end SIP model – i. e. my SIP-enabled “telephone” being able to directly address your SIPenabled “telephone”? I do agree that there is not a lot of that going on. The key question is whether most consumers will go for a bundled telephone service that covers VoIP termination as well as TDM termination for some monthly fee, or try to “go-it alone” with either SIP or Skype-based direct arrangements. Maybe I misunderstood the point you were trying to make? My basic point is that Skype is great for point-to-point communication with people you know, but SIP-based VoIP will “probably” be the winner because most people will just go for a bundled service that does everything and uses SIP internally within the Service Provider and between the SPs. system that seems the dinosaur. I always contended that ”identity” - of persons being connected, not just devices like phones - could (and should) be based simply on using virtually unique DNA of every persons as the numeric identity! 3. I can’t believe gate-waying Skype into SIP for interoperability is all that challenging. Why would it be? 4. What would/will be so hard to design and manufacture devices, which have hand held microphone and speaker or ear piece, which integrates Skype technology into it, connected either by IP line, or wireless to the line. Why need the PC or handheld? Siemens already has a USB port Skype device. Heap: I understand. I’m not trying to come across as an apologist for the telephone companies, but to a very large extent, you have the ability to talk to anyone in the world now - via a telephone. Matson: That’s not true and I keep having to remind people of this. With current day telephony, the ONLY person in the world you can talk to today is the phone company at their local central office. And only on their terms, will they then let you talk to anyone else. The power of VoIP (“my content” over dumb pipes) is that for the first time in the history of the universe, I CAN talk to anyone in the world! We ought to make sure we don’t perpetuate myths! Hughes (to Steve Heap): Steve, Peter Cohen: I’m no SIP wiz. Having used Skype, I can report that it worked like a charm across good networks. I didn’t know anyone connected in any other way. Given that a certain percentage of the population on the Internet is well connected (I’ve no clue what amount), I could certainly see offices switching to Skype type numbers and services immediately to solve their office needs and cut voice costs. What comes to mind is a country like the Netherlands that has high density of 10 million people and fiber all over the place. Offices and IP reaches ubiquity there and they could easily switch to Skype, gradually moving 1. I am not interested in voice communications to/from ONLY PSTN-provided voice telephones in the world. I AM interested in anything that provides voice communications to anyone on the face of the globe. Which, with wireless, linked into satellite, or terrestrial fiber, whether or not provided by telcos, and the Internet, bypassing totally all telephone companies. Skype seems to offer that possibility. 2. Integrating Skype into the PSTN hardly seems a big problem to me. In fact, it is the telephone numbering and code 25 Heap: Malcolm, that is a bit of a stretch - a customer has a contract with their telephone company - either a traditional wire company or a VoIP service provider - that describes the service being offered. The service being offered is an ability to make and receive telephone calls. Your argument is a bit like saying that I have a cable TV to my home but the cable TV provider doesn’t have to provide me with any channels - they simply are leasing a bit of wire to me. Matson: Steve ... You are absolutely right my argument is just like that!! I COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA know plenty of homes which have cable installed (at the homeowners expense) and it lies idle because it can ONLY be used to get stuff from the cable company and on their terms. Imagine if we had let General Motors build the roads! Sure when you say “...a customer has a contract with their telephone company - either a traditional wire company or a VoIP service provider” ...but he had no choice. We HAD to suffer that exclusive relationship with the phone company if we wanted to communicate beyond shouting distance. We have now discovered that in a digital world, making “phone calls” is NOT a service at all and relies on nothing more than the two parties having access to the necessary hardware components!! All the content is mine (and always has been) and today all I need is a piece of free software to enable me to speak with whomever I want without let or hindrance or more to the point - additional cost! The phone company grew up with a technology that dictated their presence as being essential (to allocate the scarce resource of network infrastructure) and being a state-maintained monopoly, charged the earth for doing so. At the moment, this new peer-to-peer communication (with NO phone company present) may only be possible with my neighbors across my little Wi-Fi network - but the model is scaleable to the world and as more and more user-controlled infrastructure (OPLANs) gets built, more and more people will wake up to the fact that the notion of a middle-man “operator” or “voice service provider” as we know it, is utterly obsolete .. an intermediary necessary in one technology but made extinct by the next. So I maintain as a “useful truth” the fact that in the analogue world of telephony, the ONLY person you can communicate with is the phone company. And the phone company’s only way of knowing that we have found a way of communicating directly with each other, bypassing them, will be the diminishing number of $$s on their P&L. Hughes: I don’t think it’s a ”stretch” by Malcolm, for another reason. In spite of the 1934 Telecommunications Act and its 1996 follow on, which declare voice telephony is required as a ”universal service’ and in spite of the Universal Service Fund which you pay into to ”subsidize” the phone company’s extension into rural areas - a regulatory requirement in return for their guarantee of rate of return on their investment - the cold hard fact is that ALL the RBOCs have found ways to avoid, limit, drag their feet in extending even voice telephone service across this country. Because I have worked with rural folk from the scattered towns of huge Montana (114 one-room school houses) to the poorest Hispanic counties of southern Colorado, I am fully aware of the US West saying “Sure we will extend voice service to you - the up front cost for extending the line to your premises will be $10,000 thanks...” Not data, just circuit switched and crappy voice. I wouldn’t mind if the largest telephone companies were, in fact, private businesses competing in the marketplace. But in the US they are not, and never have been. At least in some countries, including Nepal, they don’t hide the fact - the phone company IS government - a PTT. And which objects to VoIP in all its forms, for it cuts into their money stream destined for government coffers. And that government uses its police powers to prevent the competition. The only reason the Sherpas of Nepal can use VoIP is because the lowland Nepalese government gumshoes aren’t conditioned enough to climb the Himalayas and chase wirelessly connected computers any more than they can catch the Maoists. We ”pretend” Verizon is a private company, when in fact it is subsidized monopoly, where the iron fist of government is inside the velvet ”marketplace” contractual glove. Heap: All you can’t do is to reach some very remote places - although if Internet is present you can usually be assured that there is some telephony somewhere. So the issue perhaps isn’t connectivity, it is cost or quality? Costs are definitely on the way down - VoIP is seeing to that. And the prevalence of “unlimited” plans 26 that cover US/Europe and some with Asian countries is making the cost issue a little less of a major problem. Quality is probably a bit of a wash between the two systems. Interconnecting Skype to the PSTN is actually a bigger issue than it may seem - introducing a new international country code and getting every PTT in the world to program that country code into their switches and work out a way to get those calls to Skype is not a trivial problem, and if you do that for Skype, then why not Free World dial-up and anyone else who wants a private network connected? Skype could organize a big bank of US numbers and then have a second stage of dialing for the actual Skype customer, but that is pretty expensive and who pays for that? So - apart from being really neat technology that will have a big following within closed groups of people - how does it fit into the big picture of global connectivity? Coluccio: Steve, you may have touched on something more profound in one of your statements than you intended to, by prefacing with: “... although if Internet is present you can usually be assured that there is some telephony somewhere ..” Whereas, it wasn’t very long ago when that statement would have found itself standing on its own head, as in: “... although telephony is present you can usually be assured that there is some Internet somewhere.” Hughes (to Heap): I think the number of people - in the billions - who cannot be reached by affordable voice phone, is larger than you think. I deal with them, and care little for those in the urban areas of the world - for they have scads of choices. Twenty-five per-cent of the US population lives on 97% of the land area. The other 75% lives on only 3%. It took scores of years before the most basic AND expensive AND subsidized (by The COOK Report on Internet May June 2005 your phone bill, the Universal Service Charge) service arrived in all the small towns. You don’t have to apologize for the telcos. They were fine in their day. Lets honor them as we lower them into their graves - via free wireless, and a Skype sermon. <grin> moving all their (internal) voice to Skype or some other system are asking us today: Hughes: David, I am not sure what the implications are of Skype, with its proprietary protocol, ”opening up its APIs’“to 3rd party developers, as distinct from the open source code of SIP. Who owns what with Skype? And just how open will be the API access. Will Skype have “control” (royalties or restrictions) of any devices or applications using its code? The parallel to Unix vs. Windows has been drawn - now we know that security at least has become an issue in the next round of that fight - not that it s exactly been decisive, but ... how far can that parallel be drawn? Is Skype a Proprietary System Creating a Wintel-like Platform? Reed: Skype is a completely proprietary system. I have studied the behavior of Skype as I use it, so I think I know more or less how it works in gross detail. Perhaps I can even find out more. But they don’t follow any published standards above the TCP/IP layer. Their protocol for “just working” when behind a NAT box or firewall is elegant and simple, but it isn’t based on standards-based NAT traversal (such as STUN or UPnP). They don’t support SIP interoperability, though there should be no problem interconnecting to SIP if they feel they need to at a gateway. They use nonstandard presence protocols. And they exploit end-user machines, even when you aren’t making phone calls. How do you know if they are doing bad things to your machine behind your back? Just look at the active connections on your machine using netstat or whatever your OS provides. What are those connections to Japan or India doing? Are there security risks? You don’t know. Do you care? Maybe not. I tend to trust them, but all of their programmers are in Estonia, so what do you really know about what the code you downloaded does? They encrypt all the traffic, but what kind of key management is involved? Buettrich: Exactly. This is the kind of question that organizations considering What about its security? (And you may read security here in all possible ways, probably mostly as ”privacy’”). forming options that grow faster than the user base grows). Well, Skype is doing what Microsoft and Intel did in the early days of personal computers. It has created a platform that is *very* attractive to third-party developers, because of its size and ubiquity. It has invested in market share, and it is now opening interfaces and architectures that allow for others to help build value around Skype, while maintaining control of a core, and acting benevolently to those who choose to enhance the Skype platform. I guess my question is two-fold: 1. about the actual security status of Skype (Yes, it looks interesting if you capture it ...); and 2. about the value of that issue in the global marketplace. What are people’s predictions? Reed: They just recently defined and published some APIs by which third parties can use their protocols. These APIs let you do some rather nice things, and build things that go well beyond telephony. They are a lot easier to use than developing a whole SIP client, even if you have the open source code. Library APIs are much more productive than source code, even though not transparent. And remember, if you build on the APIs, you get a huge and growing installed base, for free. The base will support those APIs. This is pretty nice for someone who likes to build stuff for a market. Skype can control what it chooses to control. Of course they own their code - that’s how copyright works. They can license it on ANY terms they choose, and charge what the market will bear. They choose to let users use the binary for free, and sell Skype-Out accounts. Will they sell the later versions, or merely charge companies for the right to interconnect? They can do what they want, and will probably choose to grow their user base to create value (by Metcalfe’s Law and Reed’s Law valuations arising from the value of connectivity options and group27 This is not at all like true “open source” behavior. But it is *very* attractive to both customers and partners. Remember, in the early days, it was Apple who made the mistake of not supporting its developers. They screwed their own partners, by competing with them, or making sudden changes that disrupted its partners. They deliberately harmed both peripheral developers and software developers who got too close to Apple’s customers. On the other hand, the Unix companies (Sun, DEC, HP, IBM, ...) deliberately developed incompatible “features” that made it impossible as a third party hardware or software developer to be a Unix vendor. One had to choose which company one supported. They acted as if Microsoft was insignificant, and as if the other Unix suppliers were total enemies. I.e. they destroyed their own advantage of maturity and compatibility among themselves in the areas where compatibility and capability were advantages they “owned.” As a result, Microsoft/Intel was the truly open, binary compatible platform, and until about 1988-89, they focused on building shared value with their 3rd party hardware and software partners. IBM was the first casualty, followed by Lotus, Wordperfect, Novell,... Skype can (and probably will) play this game. They need not be truly open - they are open enough, and compatible enough. And from the user’s point of view, they are easy to use (remember the Unix guys COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA who used to say that the Unix shell script language was what users really needed? In a world of the blind, even one-eyed men seem visionary... it wasn’t hard to see that the Unix guys were clueless). And meanwhile, the SIP guys are playing the game played by the Unix players. You could never convince Sun or DEC or HP that they were in a losing game by missing the value of what Ray Noorda called “coopetition” (competition by making sure that you give enough away to partners to build the larger coalition). But they were losing, and they have lost. Now if the SIP guys realized that timing is everything, and banded together to capture users with ease-of-use and adoption, rather than trying to screw the other SIP guys in the theory that SIP is the only game, perhaps they still have a chance. This is what the Linux open source guys have figured out, as they try to re-take share back from Microsoft. If they don’t all hang together they will all hang separately. I am a total fan of open source, open standards. In a fair and clear-thinking world, open source telephony should dominate because it maximizes interoperability. But I have huge respect for smart business strategy, too. And Skype is playing its cards right, and the logical “open source” world (around SIP) of telephony is being sucked into playing its cards all wrong. Disastrously. This won’t kill Cisco, because they will get the large share of a small SIP market, and be happy. But Skype might be the Microsoft/Intel of VoIP. Join them or compete with them by changing the game. As a business decision, which would you do? I’d figure out how to buy Skype, myself. You may never be able to buy it for less than you can buy it now. Reed to Heap: I know some people at Skype, and at the moment they have no plan to allow Skype users to have phone numbers that can be reached from legacy phones. To me, that’s a plausible strategy. Remember IBM’s XT/370, which was an OS/370-compatible PC? It really did run VM/370, and VM binaries! However, in practice it was useless, because the licensing model for all of those software packages were tied to the idea of a computer that cost 100 times as much and was a corporate asset, not a personal or managerial asset. (Not to mention that all of those programs were hardly interactive.) So why take on the burden of being just a cheaper phone? Cell phones are capturing that niche more and more, as young people realize they don’t really need a home phone, and even as they become families they realize that their identity is their cell phone, not the “home phone” they don’t have. If you think about it, you can interop with cell phones more flexibly, because cell phones can call numbers that start with “*” or “#,” and if they do, they don’t have to pay access charges or use interconnect or pay all those pesky universal service obligations. Anyway, that’s enough of my free strategy consulting services for the day. point is a “service” from a service provider better than doing it yourself - $15 per month, $5 per month? Matson: David - Thanks for that very insightful piece ... it put into place lots of my own disjointed thoughts ... I am so impressed with the “user experience” of Skype compared with any of the three SIP (virtual phone) services I have tried. However, it offended my natural predilection for “openness”. Dave Hughes and I had a near faultless 30 min Skype natter yesterday between Colorado and London and we both agreed there was little to fault. But we both remarked on a slight background “swishing” noise. I have since ascertained that this was due entirely to the sound of the cash hitting the flood-gates as it was diverted away from the telcoscoffers back into Dave and my own bank accounts. We can live with that! So I guess it all hangs on the integrity, goals and degree of greed of Niklas Zennström? The problem with phone companies is that they think that telephones are essential to people, when actually they really like to connect to other people in lots of ways, and the wireline phone is merely the way that *used* to be the most convenient. For kids up to 25 or 30, IM and email now supplant almost all uses of phones, and they have never known a world without voice messaging (answering machines and voice mail exist on every phone they call). Editor’s Note: This URL posted by Matson, is an interview with Zennström on the history and current state of Skype as of early November 2004 http://www. engadget.com/entry/2635319328796286/ There will still be phones 25 years from now, but then again, there are still Western Union offices. Received a telegram or Telex lately? The reason that Skype sound is so great is that Zennstrom paid for a license to the ILDC codec. This is a Swedish codec (Zennstrom is Swedish) known also as Global IP Sound (GIPS). It allegedly cost a bundle of money. And all the other dorky VoIP services said nah. We are going to use free codecs. This is a commodity business after all. According to Richard, the ILDC codec can suffer 25% packet loss and still sound like FM radio. Heap to Reed: Don’t get me wrong - I have used Skype and like the quality. We used to use it to chat with friends in the UK, but once I moved to Lingo (and dropped my Verizon service) we find that it is so much easier picking up a wireless phone in the house and calling them under the “unlimited usage” plan than setting up a Skype connection. Maybe I am looking at this too much from a US perspective, but at what commercial 28 COOK Report: I have added Richard Shockey of Neustar to the list. Richard was co-chair of ENUM and is very SIP, Skype, VoIP knowledgeable. I had a voice conversation with Richard where I learned a couple of important things. Richard mentioned a "simple SIP" as having profiles - first a basic functionality profile, then several more advanced The COOK Report on Internet May June 2005 profiles, products could be compatible or interoperable according to profiles. SIP apparently is now embedded in a jillion IP PBXs in the enterprise and all over in carriers. But not surprisingly there are issues of control all over the place. SIP session border controllers in telco networks driving SIP folk crazy for example. one of the best European Blogs on VoIP http://eurotelcoblog.blogspot.com I will send two more posts on all this. Will SIP Win? Brad Templeton, Chairman of the Board Electronic Frontier Foundation http://www.templetons. com/brad/ Some SIP Issues Richard Stastny: Hi Gordon, good morning. Herny Sinnreich, Jon Peterson, Richard Shockey and I already had a discussion at the last IETF in Washington, DC about these issues and agreed that some action has to be taken immediately to create something like "simple SIP". Basically the current SIP standards are way too complicated for the normal user, They also they suffer from Dilbert’s ISDN syndrome. Another mayor drawback is the NAT problem, and finally the mentioned “walled garden” approach used by most carriers (fixed, mobile, cable and even Vonage-type providers) implementing SIP. As Henry always says, if you do not get a (public) URI, you do not have true VoIP. And I have to add: if you do not get a URI, you cannot use ENUM in e164. arpa. The success of ENUM is linked, one on one with open, public, standardsbased VoIP solutions. You cannot use Skype in ENUM. Since Austria is going commercial with ENUM on December 9, 2004, this is an issue for us. I agree with most statements made below and I may comment later on specific items. I finally want to add a presentation given by Brad Templeton from the EFF given at the Fall VON in the IAX2 breakout session about SIP and Skype and summarizing the rant below quite nicely. May I add another entry and name to the list: http://broadbanddaily.gigaom.com/ archives/2004/12/03/Skype-fasttrack-towhere/ This entry is from James Enck, featuring By the way, my own Blog is http://VoIPandenum.blogspot.com COOK Report: Here from Richard Stastny is the just mentioned and very good PowerPoint deck from Brad Templeton. Why the PSTN over IP (PoIP?) Toll bypass was never the answer. PoIP pleases grandpa, but is grandpa looking for a cheaper version of his phone service with quality problems? Or is it just a temporary plan for Vonage and the rest? Emulating Class 5 “IN” services not enough Think beyond the phone call Teens using phone call, even E-mail less and less. Busy signals, voice mail? “I’m going to transfer you, give me your number in case I lose you.” - Arno Penzias. Presence should stop a call from happening before it can fail What matters is the user interface, not the infrastructure. Billions spent on infrastructure, billions lost. Users don’t care how voice gets from A to B Users care about the experience. Users hate the telephone. It’s a leash. It interrupts you. They don’t know how to use fancy features. They don’t want to be more reachable PoIP is vulnerable to spam. A program can ring a million SIP phones. No way to screen based on content. Will we give up the open phone network? Will we have to screen all our calls. Not if voice is part of other applications that don’t even understand the concept of the robot caller. The robot caller is part of the old metaphor. Can Skype kill SIP? A PC application from the get-go. Did all the major things right. Easy install. NAT penetration. High bandwidth codecs. Encryption, Conferencing, P2P, scalable architecture. “Just 29 works.” Ignored open standards! Why didn’t SIP do these things? Almost all were already in specs, except P2P. People were slow to deploy, and may never deploy. Skype claims 600,000 online. How many people can you call directly with SIP? 5000 on FWD, some companies. 150K on Vonage, Packet8 almost. [Editor: these figures are sixmonths old and already quite dated.] Henning Schulzrinne: Please see http://www.cs.columbia.edu/~hgs/papers/2004/NGN.ppt for some thoughts on related topics. Interoperability is always easier if you only have to interoperate with yourself and only have a single provider, as Skype does, rather than interoperate with dozens of other companies (You typically find 60+ companies at SIPit) and work with dozens of service providers. That said, there is no excuse for how long it is taking the IETF community to come to some agreement on a simple configuration mechanism (the “Petri drafts”) that makes devices and software work out of the box. It is not hard, technically, to build a system where you only have to know your email address to get a working SIP connection. Unfortunately, most systems (that are not pre-packaged for one provider) require the user to deal with outbound proxies, several different passwords, NAT configurations and other low-level details. Doing peer-to-peer in SIP isn’t all that hard: see http://www.cs.columbia.edu/ ~library/TR-repository/reports/reports2004/cucs-044-04.pdf There’s apparently another project (SOSIMPLE) that has been pursuing similar ideas. Also, life is a lot easier if you can simply do things that would get you thrown you out of the room at the IETF - for good reasons. Skype does: - rely on a single, global authorization service COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA - transmits media and signaling on a single connection under some conditions - puts audio on port 80 (HTTP) if all else fails, thus bypassing local security policies See http://www.cs.columbia.edu/ ~library/TR-repository/reports/reports2004/cucs-039-04.pdf for an early attempt at a Skype protocol analysis. [Editor: an important paper that is really worth reading.] As identified in the VON slides Richard sent around, not being encumbered with replacing PSTN features helps. As does having a marketing mechanism (Kazaa) and not having been burnt by the first, pre-residential-broadband round of PCto-PC phones and their lack of business success. Platforms: SKYPE and SIP - Islands versus Standards-Based Cullen Jennings (Dec 27): The recent list discussion highlights why sometimes you just need to pick a phone to sort something out. I find it interesting that some times, email, particularly lists, tends to escalate in the most antagonistic of ways, phones calls are somewhat less so, and face-to-face meetings are much better. I work with development teams in many different time zones and have to deal with all the pros and cons of these various communication mediums - I think a lot about which one is not just convenient for me, but is going to result in me getting my point across to the other person in a way that I believe will be in the minimal time. I think that one of the things about SIP, is that it helps brings a bunch of disparate devices that I might use for communicating into one system. Or at least it might if we get it right. For example, I might know that I want to IM Dave Hughes because I need to send him an exact configuration string that he needs to enter in some phone in Nepal and I don’t want it messed up. On the other hand we might need a voice call so we can explain some complex issue or I might want to have a video call so I can look at the back of some IP phone he has and tell him if the poorly labeled Ethernet jack on the right or left is the one that goes to the network side. I may want to use very different devices for these communications - but I want one system. And I want one address where I can reach Dave and get to the device he wants to use. I want to know what types off communication he is willing to receive at this point of time. I don’t know if he is sleeping in Nepal or wide awake in Colorado. Should I ring his cell phone and wake him up? Or should I check his presence, IM him, then escalate to a voice or video call. Now Dave might have a bunch of different devices that get connectivity in a wide variety of ways. And when we are talking about Dave I should probably expand that to a *very wide* variety of ways. Dave is going to get those devices from a wide variety of vendors, he is going to get the connectivity from more sources, and then he is going to expect services like messages storage for IM, voice, video, and conferencing and collaboration tools from another set of providers. So, I’m sitting here on my Mac staring at two applications that are basically softphones. One is called xten and the other Skype. What’s the difference between them? Also note I have advanced Beta of both of these so the features I am seeing might be slightly ahead. Let me list some things that I think are critical to both of their successes: - They are free to end-users. - They work on a wide variety of platforms and computers with a high install success rate. They sound good - the audio quality, particularly with packet loss, is good. This is not surprising - they both basically use audio technology from GIPS. - They work behind a wide variety of NAT and Firewall scenarios. They both basically use a NAT detection technology that follows a model called STUN plus fallback to TCP tunneling. 30 - They both do IM and presence related things. - They both do audio and xten does video but I cannot imagine any reason that I will not end up with video in both of them in a fairly short time. When I ask others, I often get told some of the things above - I also get told about one is a P2P system and other is not - I’ll save this for a different thread but I’ll give you a hint that I believe that both of them are P2P systems - in all meanings of the term P2P. So, I’m left with what is the difference - one of the differences is that one of them has a name that rhymes with hype. So let me go back to the scenario of talking to Dave. It was about multiple devices that acted like one system. I want to send him a message at one address - and he wants to have many devices that act like one system. Now one easy way to do this would be to have many devices, that all came from one manufacture, and were all part of one system and I could only talk to people in that one system. A long time ago, I used to have an email system that worked this way - I think it was called CompuServe. It was a major player - which was just proved by the fact that my spell checker has the word CompuServe in it’s dictionary. Now, this was no doubt good for CompuServe, but it was not good for Dave. (I don’t know but I’d bet a beer that Dave once had a CompuServe account and no longer does). Why was it not good - well basically new players could not connect and provide new services - particularly ones that competed with the heart of how CompuServe made money. Which of course leads to the interesting questions of how Skype makes money because this will determine what can and cannot be done with it by third parties. People asked for a standards-based way to do email. Now when I send this email, I have no idea how it will goes from Gordon’s email server to his email reader. It might be IMAP, It might be POP. It might be some proprietary protocol from an exchange server. I don’t know and I don’t The COOK Report on Internet May June 2005 care. But I do care that my system is going to send his system an email and we are going to agree on one address to use and one protocol called SMTP. So what will the Skype island use to connect off the island? Or will it? I notice that the PSTN connectivity seems to come from translating Skype to SIP and terminating on a SIP gateway. I find the islands game silly. Right now we have a few major islands of IM Presence - AOL, MSN, Yahoo. But I predict these will move to standards-based interoperability or slowly slide into oblivion. Think about the PSTN and the common carrier concept - can you imagine having a MCI, ATT and SBC phone on your desk and you need to use the right one to call someone else because you can only call on the same network. People will do this if there is no alternative - but there is an alternative. So let’s drop down a level of things that concern me and things that I don’t know what Skype does. Security: SIP has potential for encrypted communications between people who have never spoken before with a deployment model that does not require any extra effort or cost on the end-user’s part. We have an Identity scheme that can be deployed and is a major part of dealing with SPAM. Ask me some other time if I think that stored IM could someday replace email. Skype has security but won’t tell me how it works so I can’t really comment on it other than the comment I just made which is about the most damning statement I could make about any security system. Multi-device, multimedia support. If I have several Skype devices all with different capabilities and some on and some off, what happens when Dave calls me. Does one device ring, all of them, just the right ones? What happens if I answer two of them? Some of the key capabilities of SIP are wrapped up in this area. What is Skype’s business model? Not surprisingly, they don’t seem real keen to tell me about this. If the business model is to have traffic on their network that provably has nothing to do with file share as a tactic to stop certain sort of court orders - well rock on, I get it. If it is they are going to make money selling PSTN termination, well I note there are a lot of people doing this and I’m unclear what Skype’s sustainable competitive advantage is. A few years back Microsoft released a voice.net system that allowed MSN messenger to PSTN calls. Most the TSP that provided that used Cisco equipment but there is not exactly a horde braking down the gates to get more people on this voice.net system. There are a ton of people breaking down the gates to provide VoIP services but they mostly seem to involve many services and systems that interconnect in complex ways. Some things I would like to understand: 1) What are the differences between Skype and SIP Softphones? 2) What is Skype’s business model? 3) Will Skype work with many devices over many access mediums and get the right device, at the right time, with the right media? So where do I stand on all this? Skype has got my attention - it’s a nice application of VoIP with a good adoption curve. I want to understand why - is it a fad and hype, or are there some key differentiators here that really make it different? As people who know me will note I often do, I’m going to show up and watch Skype and SIP, pay attention, tell the truth, and not be married to the results. This is some pretty unstructured rambling to send to a group of people like this list. Part of why I joined this group was to help listen to what others are saying and refine my views on Skype. Reed: I have spent my career arguing for open systems like SIP was supposed to be, not open systems like “Unix” turned out to be. I was going to write a long comment, but instead let me use a comment that I think is due to Peter Drucker (at least it reflects his way of thinking). The way to succeed in business is to pick 31 the best customers, and delight them. And the crucial caveat - the best customers are not the ones who always buy anything you sell - those are *your* best customers, not *the* best customers. The best customers are the ones who will teach you what you should be selling. The following is how it applies here: SIP’s vendors have defined their customers to be phone companies. Skype has defined its customers to be people who live a communications-centered life. It’s impossible to delight a phone company with voice over the Internet. The people who live a communications-centered life will teach you what really matters. Those people are *not* happy customers of the phone company. It’s still possible to beat Skype with SIP, but the current SIP vendors (such as XTen) have no clue whatsoever! To win, you have to delight some customers, not participate in an illusory “market” for “technology”. Here’s why Dave Hughes matters. He’s delighted! I’m sure he’d be even more delighted with a truly open system. But he can’t get an open system, with full interoperability, today. And he can’t because the SIP people think the game is about making the incumbent phone companies happy. So openness (as defined by market-enhancing and market-delighting interoperability and ease of use in many conditions) plays second to controlling the users. That’s what happens when you have MBAs run your company instead of true entrepreneurs. Open standards need both entrepreneurs and innovative technology. I’m not sure which is more important. Microsoft got where it is by entrepreneurship, and a slight bit of open APIs (partly contributed by IBM’s PC architecture choice). X.500 was “open” but controlled by phone companies (well the OSI, which is a consortium of phone-thinking people). COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Internet mail headers were “open” and entrepreneur-led. It all starts with the endpoints, and delighted users trump slow-moving phone companies. When Pulver started FWD, I thought there was a chance for SIP. But when the SIP vendors didn’t embrace it enthusiastically, that revealed FAR MORE about how the vendors think. I remember sitting in a room in the early 1980’s with very senior IBM executives explaining to me patiently how “real personal computing” was what you did on a VM/370 machine with a 3278 terminal, and that once we ported 1-2-3 to run on VM over 3270 terminals, we would see enormous revenue growth. They believed this so much that they were willing to, and did, fund the complete development of 1-2-3 for Mainframes. I cannot tell you what the sales of that product were. But I never met an actual customer - much less a delighted one. We at Lotus made a reasonable return on our development $. But it would be hard today for me to argue that it wasn’t a complete strategic waste of time for both companies. If SIP is a good technology, then I would ask why Skype found it necessary to use something else? And if I were a SIP vendor, I’d be going hell-bent-for-leather to sell my solutions to Skype. And if I were a SIP standards person, I would get off my fat a**, stop enjoying the free travel that seems to be the perk of going to all those IETF meetings, and start figuring out why my standard sucks. SIP Must Interoperate While Skype Need not Do So Schulzrinne: Not being an open standard has a few advantages: - You can limit yourself to working with only one Internet-to-PSTN “carrier” (Skype) and one, global authentication server; many of the problems experienced by SIP users are configuration problems that are much harder to solve if there are multiple service providers. I readily admit that the IETF community has not nearly paid enough attention to configuration and diagnostics issues. - You can do things that would get you in trouble with corporate security folks or any network engineering group, such as running voice and call control over port 80 (HTTP) to evade firewall restrictions or running both call control and voice on a single transport association. - You don’t have to worry about interoperability as you control all the software. - You can do security by obscurity. The success of non-open applications in general, probably has little to do with technical superiority, but it does point out the real costs of open systems in terms of testing, brittleness in the field, customer configuration and the like. The Web had it a bit easier, with essentially one or two dominant pieces of software, on both the browser side (Netscape and IE) and the server side (Apache and IIS), but even there, the experience has often been frustrating. Continued dominance of proprietary applications, as shown by IE (ActiveX), also often has little to do with technical superiority but rather, with bundling advantages. Skype and Microsoft certainly enjoyed that advantage. Neither FWD nor iptel.org nor all the other providers of similar PC-to-PC services had anywhere near the name recognition and the ability to cross-market. (I also suspect that VCs, having “learned” the lesson of earlier PC-to-PC VoIP failures, ignored the fact that the deployment of broadband made this much more viable than attempting to do voice-over-modem.) I’d be curious what you would cite as evidence in SIP of the sinister influence of PSTN types. Yes, there are things like “early media” that are influenced by the need to interoperate with legacy systems, but they are very much at the margins and don’t really shape the overall system. 32 Reed: Sinister is your word not mine. I don’t think it is sinister. It’s just sad to waste time pursuing those customers. I can tell you that many of the SIP vendors are spending lots of time with those customers. And they build their business plans based on numbers from those customers. One piece of evidence is the battle over ENUM, as if it matters. Another piece of evidence is the lack of urgency in finding a way to tunnel SIP over NATs. (I’ve watched the STUN process and I can tell you it’s been a travesty - why does Skype work over NATs today? Why does STUN screw up?). Of course you could blame it on stupid engineers compared to those at Skype. But I think the bias has been that no one feels that working over NATs is important to SIP’s success *in the near term.* Where’s the urgency? This standards committee views itself as having years, because they can’t imagine any users other than phone companies, and they know the phone companies feel no urgency. Jennings: Hmm, Cisco has been shipping SIP VoIP stuff that works over NATs for several years now. AT&T, Vonage, and others have a fair number of customers with Cisco/Linksys VoIP devices behind NATs. SBC companies such as Jasomi make it easy to use stuff not designed to work behind a NAT work behind a NAT and they do not deploy any equipment on the customer premise but only a central relay server in the network. Is there a lack of urgency or is it a solved problem? I’d actually say that there is a fair amount of urgency and it is a partially solved problem. Reed: I’ve used ATA-186’s. When they work out of the box, they’re fine. But when they don’t “just work” (which is much of the time), the user either returns them to Vonage, or spends Vonage’s customer support time beyond any reasonable profit for the next 5 years. My point is that the service needs to “just work.” I commend the Cisco product The COOK Report on Internet May June 2005 designer to look at this UI for setting up and diagnosing your typical SIP interface hardware: [Editor: David Reed includes a screenshot of a complex, cluttered and arcane looking table.] Jennings: Completely agreement that we need to get to where things just work out of the box with no messing around with configuration. The ATA was pretty much our first generation attempt device and is around 4 years old now. I was referring more to the Linksys stuff that came out almost a year ago now - like the Linksys PAP5. TSP can order this preconfigured to their servers and I believe the just-work-out-the-box success rate has been very high. No doubt additional improvements are possible. Reed: Cullen, thanks for the info - I’ll look at the PAP5, but I suspect it’s very likely that it won’t work out of the box (for any TSP) at my MIT or HP offices, to pick two very different but highly structured environments with very different firewall policies. Whether it would work at a Hotel or my home behind a Netgear RP614 NAT box, that’s more likely, but still not a lock. If it just works in all places, that’s very cool - which VoIP provider ships it? Skype, of course, works without my needing to think or intervene in all the above, including over my HP laptop when it is connected to the HP VPN, which bars split tunneling, so Skype is going through the VPN. And this comment only refers to the tunneling/STUN aspect. A number of the SIP vendors seem to be waiting for, and expecting to require, QoS solutions in the network before entering the market and learning what really delights customers. Perhaps they believe that QoS is required to get acceptable performance. (After all, if you don’t actually use the network, but instead spend your time on airplanes and in executive suites like many of the decision makers in major networking companies, including my own part-time employer HP*, you might believe your own gas that QoS is needed, as opposed to being a nice to have. Remember call-waiting required ISDN... until it didn’t). *HP Labs employs people like me (former CTO’s who still spend 50% of their time writing code and operating spectrum analyzers rather than turning into pundits, which is much more lucrative but less fun) to remind them that technology reality is not discoverable at Industry schmoozefests full of powerpoint, market strategists, and “glad to be here” commentators. Needless to say, I do not speak *for* HP in any way here, though I sometimes speak *in* HP in exactly this way. What to Do About NATs and Protocol Problems? Earlier Jennings: Lots of people have played with tunneling audio and video over TCP. At first they are thrilled that it worked though a firewall that blocked UDP. But somewhat later, they start complaining about voice quality problem - The complaint sounds like” ”It’s fine most of the time but sometimes it just drops out for a few seconds.” The issue is TCP slow start can cause a 50 packet per second stream to back up significantly when two packets are lost. Then people move to a stage of thinking: “Ah, let’s replace the TCP stack so that the stack ACKs stuff if never received.” Ignoring what this bad idea would do to the Internet in the large, my experience in trying to deploy software to replace the windows TCP stack is that it would not be fun. It’s very difficult to diagnose, replicate, or fix, the voice quality issues that happen over TCP tunnels. Reed: I think you are confused if you think this is the problem with sending voice through firewalls when the capacity of the firewall is sufficient for voice in the first place. In such a case, you merely need to keep the flow of bits going at a moderate rate, and slow start never kicks in. But that’s being a pragmatic engineer, and looking at the problem as a simple one, rather than trying to optimize an objective function (a few wasted bits) that means nothing in practice in a world of fiber, coax, and capacity that is much cheaper than Verizon attempts to extort 33 for POTS bits. Of course the firewall guys basically are sticking a major barrier into the pipes, and you can’t do anything about a constriction in the pipe that makes it so the load exceeds the capacity. That’s why I describe the people who build firewalls as people who believe you can improve highways by standing police in the middle of the highway to direct traffic. Jennings: I think a better path is to take a balanced approach of fixing the network such that real time P2P applications can work well on it while at the same time being realistic about the deployment of IPv6 and NATs. Some guy called D.P. Reed has this great paper about the “End to End” principle that I strongly believe in :-) To do this, I have been working with the NAT and Firewall vendors. For all intensive purposes there are 3 vendors of home NATs and threee vendors of non-home NATs so it’s not that hard to reach them all. The goal is to make endto end-possible but still allow administrators of firewalls to impose the policy they wish. If they want to stop a certain type of traffic, they always can do it so the goal is to make it easy for them to do and make it so that they don’t accidentally kill other things in the process. Reed: Most of the network works fine already, on a performance basis. It’s not clear we need to add stuff to create QoS. Perhaps we should delete some stuff (like firewalls). I run real-time collaborative p2p applications (Croquet) between my home in Boston and Cary, NC; Magdeburg, Germany (former east germany); and Palo Alto. Often I use hotel networks, too. The latency and jitter is quite acceptable. But getting the firewall and NAT vendors to realize that the *illusion* of safety and control they produce for their customers is killing the value, and creating none in return. Spend half the energy and investment on deploying security at the endpoints, and enabling rapid fault and intrusion isolation at the network layer, rather than trying to prevent faults and attacks. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA The Value of Interconnected Networks COOK Report: Here is what I have heard so far from Skype. Kelley Larabee: Hello, Gordon -Thanks for contacting us. Skype already interoperates with PSTN, and we have announced that we’ll offer SkypeIN as a premium services in 2005, so that Skype users can be contacted from the PSTN network. We expect that as Internet Telephony grows, so will cross platform communication demand and potential. Our considerations for evaluating other systems will include: Quality of service (call set up, audio) Privacy (encryption or not) Ease of use Cost (server based or P2P) Enhanced services (IM, presence, video, file transfer, etc) Coluccio: I found Skype’s comments about their already being interoperable with the PSTN interesting on several levels. Some of the VoIP box vendors I’ve spoken with, whose primary businesses are in session border control and gatewaying functions, appear to be almost entirely oblivious of those powers (some are entirely), almost straining themselves to recognize the name Skype, itself. Much less see the need to regard them as a real threat, or the need to interoperate with them among the various flavors of H.323 and SIP that prevail, which they now support. Then again, I really need to get my head out of the trappings of carrier-enterprise quarters more often. Vijay Gill: Skype has SkypeOut, which works somewhat and unfortunately, isn’t the cheapest thing around. I have been using Skype with great suc- cess over my Verizon EVDO connection, allowing me essentially unlimited calling for a flat rate, that is also mobile. I use SkypeOut to call the PSTN for rates cheaper than my average anytime cellphone minutes, and obviously other people on Skype work just fine. SkypeOut rates to India for example, are 0.15EUR/min, which is more expensive than the rates from Gorillamobile at 0.14USD/min. The Skype codecs and user experience are very good for a person who just wants stuff to work. Hell, Skype even manages to work decently over a GPRS connection via Bluetooth to my GSM phone, and GPRS latency is about three to four times that of EVDO. Jennings: If anyone is going to be at CES next week, might want to check out what Jeff Pulver is up to with his FWD stuff and Skype. h t t p : / / p u l v e r b l o g . p u l v e r. c o m / a r chives/001484.html In my conversations with AOL, MSN, Yahoo about services that allow messages on one system to reach users in another system - they have all been fairly clear that, other than very special conditions, this violates the End User Licensing Agreement. Reed: This ranks up there with suing your best customers. Shockey: Where did you hear this Cullen? MS is bridging all three now using Microsoft’s LCS 2005 http://www.internetweek.com/allStories/ showArticle.jhtml?articleID=23901035 Jennings: Microsoft’s bridge allows a Microsoft endpoint that is paying money to Yahoo and AOL talk to a user in AOL *or* talk to a user in Yahoo. This makes sense. AOL can generate revenue off this. Microsoft’s bridge does not allow a Yahoo user to send a message to the bridge that gets translated and delivered to an AOL user. If this happened, there would be no 34 stickiness to existing AOL users to stay with AOL because they could switch to jabber and bridge to their previous AOL buddies. A model like this has no revenue in it for AOL and lets AOL loose all their customers - phone any of the public IM vendors and say you want to deploy a bridge that lets any user talk to any user and see how they react. Reed: Not that it matters, but both Hal Varian and I have pointed out that the economic option value created by interconnecting separate networks far outweighs the value retained by balkanizing communications against the user’s interests. The only case where this doesn’t work that way is if one of the parties believes they can capture a monopoly position, and then use that monopoly position’s uniqueness to extract “monopoly premiums”. Once upon a time, AOL might have had a shot at monopoly construction in the IM space. Now that we have P2P-implemented IM (e.g. Skype, among many others) the rationale for an IM operator to exist declines rapidly. The same transition from conceiving a service as centrally provided to decentralized, user-constructed, is why email (and before it telephony) saw no economic benefit to retaining balkanization. Or for that matter, why zero-cost peering grew. The mutual gains to both sides of a gateway are greater than the losses due to loss of control. There’s a lot more theory and subtlety here, but IMHO, AOL’s knee-jerk policy has outlived its usefulness to AOL. If it wants to survive, it should figure out how to ride the commoditization curve of IM while maximizing its profits, rather than sticking to a strategy that will kill its own golden goose. Jennings: I would like to build a conference bridge that allows Alice to connect to it with MSN and Bob to connect to it with Yahoo. This violates both of their EULA. Reed: The idea that a EULA can prevent all functional substitutes from competing is silly. If you don’t realize that IM is a trivial app, and AOL users do have a choice to switch, you deserve to lose. The COOK Report on Internet May June 2005 Jennings: One of the best people to explain the complex legalities of this is Jonathan Christensen who was at Microsoft but is now is CTO at Facetime that specializes in conversion from one IM protocol to another. Shockey: Nothing like the ILBC codec that is the best in the business. Now Skype did have to license that code for real money and you can’t get it in other products like XTEN unless you pay for the licensed version of the product. Reed: Complex “legalities” are why lawyers shouldn’t pretend to understand business. The more complex the deal, the more likely it’s just “security by obscurity,” making the parties to it incredibly vulnerable. The best deals codify the interests of ALL parties in a joint win. Writing a EULA to screw your customers is VERY bad business, and only works if you have a monopoly position. http://www.globalipsound.com/ Jennings: You - David P. Reed wrote “And if I were a SIP standards person, I would get off my fat a**, stop enjoying the free travel that seems to be the perk of going to all those IETF meetings, and start figuring out why my standard sucks.” What the Technology Must Do in Order to Please the User Jennings: I’m all ears on this - I’m very interested in what to improve. If standards don’t matter, then we need to get the meetings moved from Minneapolis to somewhere nice and warm :-) You mention that people really like Skype and that it delights real users. Can you dig into that a bit and get specific - What is it about Skype that people really like? Reed: I’l let other users like Dave Hughes chime in. No particular order, except #1 is by far the first among equals. 1. It just works out of the box in minutes, no matter where you are (zero configuration setup). Shockey: Yep. It just works. Even a certifiable SIP bigot such as myself uses it all the time. Reed: 2. The voice quality is excellent. Reed: 3. Conferencing is trivial to do, and free. Shockey: Yes, but Skype had the luxury of vertically integrating all the various elements of a coherent VoIP system without having to interoperate with other implementations as SIP must. Reed: 4. No phone numbers just names, you start with “presence”, and typically use the IM feature to ask politely, with a topic in mind, if the other person wants to talk now or later. Shockey: Ditto but that is a User Interface issue that now everyone is going to mimic. Reed: 5. No “call waiting” - you can blend multiple IMs, voice calls, conferences, ... with a nice User Interface that exploits it. 6. You don’t have to hold a telephone in your hand when you want to use a mouse and keyboard to do other stuff. Shockey: But you can do that with any decent softphone such as XTEN Reed: 7. And of course, it’s free (or rather, since you are already paying the cost for Internet access, and Skype uses such a trivial amount of extra resources after that, it’s a rounding error). 8. Finally (this was a bit late), Skype just works across OS X and Windows, and a few less popular platforms. In contrast, every SIP user experience design I have seen (such as XTen) is modeled after the POTS User Iinterface that evolved based on the restrictions of a central, legacy switch, with dedicated circuits, 10 key pad, special case conferenc35 ing. And every SIP install is a nightmare of “settings” and failures to operate over various firewalls, tunnels, etc. Admittedly, you could do everything that Skype does based on SIP. But to do so, you’d have to admit that telephony as we know it sucks and is worth redoing all over again to make it fun and productive. Heck, maybe the generality of SIP might even be exploited so that you can get a voice bandwidth of 1616KHz, and a dynamic range of 96 dB, rather than the typical 3 kHz/8 bit audio that is the best one can get from “toll quality” sound. I could play my flute for my daughter when I’m on the road. My point is that the *potential* for delight *is* hiding there in SIP. But the folks who do SIP have invested no time thinking about it, much less time making it happen. Stop hiring engineers and people who think writing standards is their idea of a great way to spend a weekend, and start hiring creative types who have strong views of how cool things could be. XTen, for example, isn’t cool - it’s just a picture of a phone on my screen. Menard: To be fair here, you have to bring in Robert Sparks, CTO of Xten into the picture. http://sipthat.com/archives/000143.html I think that their Eyebeam product with support for H.264 video and Speex (Editor: a free voice codec used in Windows)Wideband audio (http://www. speex.org/comparison.html) is setting the standard for the implementation of such functionality in IP Phones: http://www.xten.com/index.php?menu= products&smenu=eyebeam&ssmenu=e yebeam-sdk If I want to call a PSTN number, I like to have a keypad there, but with ENUM support, I can easily as well punch in an email address-SIP URI. I hope that IP phone vendors will agree to standardize on a PS2 connector for hooking up a full QWERTY keyboard to the telephone. I’ve been trying to convince Mediatrix to implement Speex WB into their 2102 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA without success to date. http://www.mediatrix.com/products_devices.php?prodid=14 Which I am told has a wideband capable SLIC and a 12-bit A/D-D/A therefore capable of sampling much wider bandwidth than 3.2 kHz at a much better sampling rate than 8kHz. What we need are new wideband analog phones. Curiously enough, the cheaper the phone, the more wideband it is. So David’s quest for a better audio quality than POTS is echoed by mine and is quickly going to face the ugly battle of inter-carrier interconnection using something better than G.711, which I have raised on the record of CRTC Public Notice 2004-2 and is the subject of the CRTC CISC NTWG TIF 14 on VoIP (http://www.crtc.gc.ca/cisc/COMMITTE/N-docs/NTTF014.doc) which was just reborn, but is not drawing contributions until the CRTC sets policy directives as a result of PN 2004-2 on VoIP. Boy would I like to dump better than POTS audio on Bell Canada’s network, but I am certain that unless I battle it out in front of the CRTC, Bell will insist on transcoding this audio to G.711. Now, to assume that there can be a public Internet that replaces the PSTN using Skype as a the user interface, doesn’t imply interoperability with the PSTN, which we can toss away; but is still where 80% of the calls will be originating from and being placed to for a long time to come. Dave Hughes: Well I have only intermittently been able to follow this thread and haven’t commented, Xmas, kids, grandkids, and great-grandkids ya know. But I will answer this question by Cullen “What is it about Skype that people really like?” And David Reed suggested I comment. Ok, I will. I have been using/evaluating Skype for a while, getting other BUSY people to use it, pushing it to and getting calls back from the most arcane and difficult places. And I have been using and trying a range of other solutions, over broadband cable, DSL, dedicated T-1, wireless, satellite. Domestic and foreign. From high quality Cisco 7960 phones with tweaked codecs, Vonage with AT186, both here in Colorado, in hotels with IP room service, and by carrying the SIP AT to Europe and connecting it to a local broadband DSL then making calls to PSTN phones in 13 Third World countries from Senegal to Bhutan, MSN, Pulvers FWD, on Win2000, XP, variation of FWD ported by a couple of Russians to an HP Ipaq, itself connected to Wi-Fi (would have exercised Lingo on a trip to Denmark but their management sucks - even though their engineers are ok). So I have some basis for my comments. 1. Far and away the biggest appeal to Skype is its incredible voice quality. Commented on by everyone when they first hear it. Under a range of conditions of bandwidth. In fact it is much better than PSTN phone calls, local or long distance. The only time I have heard PSTN calls as good is in corporate board rooms with the highest quality equipment, room speaker or handsets over excellent circuits. In fact at its best - between two PCs connected to cable/DSL level bandwidth, its almost eerie. As if you are in the same room talking across a table - without background noise. In fact I would say it’s better than a face-to-face conversation with its sound distractions! And I’m half deaf at 76 years old from war battering of my eardrums. I would use a set of ordinary earphones into the audio port on my computers talking over Skype before using my $4000 (each) digital, tunable hearing aids. International calls on SKYPE (Japan, Europe) are just as good quality as PSTN and better than Vonage. That’s 75% of it. All the rest are of lesser import. And there are some downsides. 2. Yes, trivially easy to download and set up and make the first call. Smart enough to get through most home/of36 fice user routers without configuration. Only time I had a problem from a 2000 laptop was after I ran a multi-media CD on the system and it “reset” the default voice/mic setting in the Windows Sound utility - so the mike outgoing didn’t work after it had worked. Until I reset the card to default. 3. Works at the amazing low bandwidth - where even the lower speed codecs for SIP fail. I have had successful calls from Ghana, Africa over a pisspoor PPP dialup connection at the other end calling me, across the capital city into a local ISP service with hardly a high bandwidth transatlantic connection. And even though, in Thame, Nepal at the end of four relay Wi-Fi radio hops 5 miles, from/to a geo satellite link between Namche and Kathmandu, operated by marginally tech Sherpas, and a SECOND geo satellite hop from KM to Hawaii and THEN to the continental US into the network where the Namche sat hop is only 64kbps and the total latency is 1,200 ms (about 88,000 miles up/down) - or 7 full seconds turnaround Skype STILL worked well enough to catch voice inflections and quality. Telling me I can reach ANY spot on the globe, voice. 4. The easy “name handle’” + full name in the scan/search is simple and better than numbers, though because geeks like to use cute names which are not memorable, that’s a downside too. Now whether Skype works better than SIP ought to work is, as Dave Reed says, it got waylaid by the bell heads trying to turn it into the old PSTN model, and thus got so encrusted with features and barnacles of code, or some other reason, such as just one hell of a superior codec in Skype, it sure is giving SIP a run for its money. I’ll stop there. But I want to repeat what I said to a large (200) 32 nation conference in Denmark of wireless ISPs or wanna bees half from very Third World countries - even before I had Skype running - that I think the #1 killer AP for the REST of the world can be VoIP over Wireless the last 100 miles. BECAUSE in nations like Senegal with only 7% literacy, the ONE thing EVERYONE can The COOK Report on Internet May June 2005 do is speak and listen over a “telephone.” THAT should come BEFORE computers to the jungle. Because if you have a small, cheap, Skype phone set with a wireless link you can TALK THROUGH anyone staring at a computer screen how to operate it. While you can’t use email to teach them if they are illiterate in the first place. That makes a market of about 6 billion by my reckoning. Jennings: Thanks to Dave and David for the delightful lists - I heartily agree those are very key features for a good system. Hughes: Yeah, Cullen, I felt bad crowing about how Skype worked, while the Cisco SIP 7960 phones stopped working when the satcom bandwidth to which they were connected dropped to 64kbps, to the most nether places on Mt Everest (Thame, Nepal) where voice - to teach English to young Sherpas whose Nepalese teachers can’t even speak/write it - is critical. After Cisco via Jim Forster and you were so generous with your devices and time setting them up, And troubleshooting a DSL firewall problem Mingma Sherpa in Pittsburgh was having. Had another Skype “voice quality” demo when Frank Coluccio called his first time. His deep booming voice came through splendidly. He had a small hardware problem at the end, which illustrates one of the problems any PC-mic-speaker VoIP runs into sooner or later. Others including me, have encountered it. Whether its Skype or Pulvers FWD or whatever, the arcane and very seldom accessed - by the average PC owner – “Sound and Multimedia” Control Panel settings for microphone or speaker/earphones have to be fiddled with if a user - or software - has altered the default settings that Skype expects. Most often it is the microphone - at one end or the other which has to be fiddled with and reset before things work as advertised. The issue here is - how many “average PC owners” know or care how to go down five levels to Start - Control Panel- Sound & Multimedia -Audio - Preferred Devices then see such terms as “Microsoft GS Wavetable SW Synth” or “ESS Maestro” to change settings that their music software reset when they installed it? And then know how to test it? Five of the 10 or so tech hot shots I first communicated with with Skype, had to fiddle before their microphones, or mine worked right. Monitor screen interface easy, installation and getting through Cable or DSL modems easy. If mic and speaker settings are not default right hard, or frustrating. If Skype is to take over the world, its got to be embedded in hand held telephone type devices, where even setting time/ date, voice answering, call lists from an instrument’s LCD and punch buttons is a turnoff to many who just want to pick up a familiar instrument - called a telephone – “dial” a number (name?) and say: “Can you hear me now?” Not sure PC screen based VoIP will subsume all SIP ”telephones” and/or AT devices. Forster: Yeah, I agree. This is one place where the Skype experience is sometimes not great -- the microphone/speaker end of audio, where many phone appliances do “just work”, and the complexity of interaction with other apps. We could pick nits and say that’s not Skype’s issue, it’s a PC platform issue, but that’s similar to saying that the firewall/NAT issues aren’t SIP issues. So far we have VoIP phone appliances with ”just works’”audio, and Skype with “just works” setup. One of these days we’ll get both in the same device. I think I’d bet on “just works” ease of setup in a phone appliance before “just works” audio in a PC. Changing Expectations of What People Want from Telephony The question is why not do better than PSTN from the get go. Reed: Francois - I want to correct your comment that 80% of all phone calls with be originated on POTS for a very long 37 time. That’s not the way the market is going if you look at the big picture demographics beyond just the IP world. 80% of voice will originate and terminate, not on POTS, but on portable phones that people carry with them and connect wirelessly. Don’t confuse them with POTS - they are a different architecture altogether. (Consider the push-totalk feature as a bellwether of how nonPOTS services drive new markets). Menard: In Canada, the market is now controlled by three players, Bell Canada, TELUS and Rogers, all incumbents, all of which have no short term commercial interest in simplifying the terms and conditions of interconnection unless they are forced to. I fully draw a parallel between POTS and PCS. Reed: This is a documented market phenomenon that is growing along with the age cohort who is comfortable living with cell phones as their primary phone gets older - at least one year per year, probably 5 years per elapsed year. Probably even a higher percent share of calls if you are referring to phone calls that connect person to person (as opposed to terminating in a recording device to leave or retrieve a message), since reaching a person is far more likely on a phone that is with you rather than where you live or sit at a desk. (I’ve been trying to get someone to measure recorder termination for about 15 years now, because it fits Nicholas’s model that async communications is one of the preferred modes for users, but we don’t design our telephony systems around async recording transport). Breaking the culture limitation on VoIP growth now becomes just a software problem: because these telephone endpoints are smart, the network can be application-ignorant, as it should be to foment innovation. It would be easy for Qualcomm and Nokia, for example, to add Skype (or for that matter SIP) to their phones - they already have IP in them, and much better User Interface capabilities than do the stupid wired phones. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Menard: I do not agree. Breaking the culture limitation on VoIP growth requires the telecommunications regulators to do their job and implement unbundled access to ILEC DSL and MSO Packetcable loops at competitors rates that are cost-based and limited to a reasonable mark-up and which do not systematically price squeeze competitors. Once this is in place, people will implement VoIP using alway-on, IP-to-IP user interfaces, which will inteed, relegate E.164 phone numbers to their grave. I foresee the buddy list interface and use of e-mail addresses as proxy to SIP URIs for example. However, fact remains that people on PCS phones and on POTS phones will have a limited user agent and user interfaces for a long time (3 years replacement cycle for PCS phones that are crosssubsidized by long term contracts is not uncommon) and will be constrained to a 12 digit keypad focused on E.164 entry and audio that is too thin for effective voice recognition. By contrast, VoIP users may take advantage of better user agent hardware with wideband audio capable handset and a QWERTY keyboard, which are problems that can be cured, not by software, but by open standards and terminals that are not cross-subsidized by a network operator oligopoly. For instance, VoIP providers are not advertising BYO SIP UA (Bring your own SIP UA). This is a sharp contrast with the PCS operator oligopoly. The point here is that with VoIP, those things that cannot be cured by software, can be cured by improving the computer that talks IP. With VoIP, there are no preconceived ideas of locking the user out of its own user agent (such as disabling third party call control on all Windows CE-enabled CDMA phones), to the point where you cannot even build yourself an applet that tracks the time you spend on the phone. Reed: The idea of phone numbers is only going to be relevant for a short time into the future. So don’t bet on ENUM being a “cash register” for very long at all. Menard: ENUM will not be a cash register. It will be a means of forcing SS7 operators to DIP into the Internet rather than the converse which is currently the case. I alluded to ENUM as a mean of hiding the complexity of PSTN call routing from a user interface that is backward compatible with the POTS and PCS user interfaces that are deployed today. Reed: When I can beam my Skype ID from my phone to yours when we meet (instead of giving you a piece of cardstock with tiny printing, or in a bar having you write it on a napkin), when we have online directory services with voice recognition interfaces, why will we ever need numbers again? Many people are switching away from the location-based notion of telephony to the person-based notion of telephony. Just like “answering machines” went from being viewed as annoying to have, to being annoying when someone didn’t have one, not having a Skype name (or something better) will soon become a social problem. “Oh, HE still uses a phone number” will be the ultimate put down. Menard: The opportunity that I see for ISPs is to become CLECs so that they can elevate themselves to a regulatory stature that is that of a PEER OF EQUAL standing with the ILECs in the eyes of the regulator, despite being recognized as financially incapable of reaching the scale of the ILEC in any short amount of time. Today, ISPs cannot peer with ILECs at terms and conditions that are non-discriminatory. Say I’m a small ISP in Trois-Rivieres, I do not want to go peer with Teleglobe at the NYC NAP to send traffic back to Bell Canada. If I’m a CLEC, I can interconnect locally at the LIR Central Office (LIR = Larger Interconnection Region as defined in CRTC Decision 2004-46). I’m afraid ILECs will soon argue: Oh, but wait, this is not a VoIP packet, its an IP packet that doesn’t contain voice, I cannot route this packet locally on my network, you have to go to the NYC NAP to terminate BGP peering ... this just isn’t the way we route our IP network ... 38 I’m making an effort to make sure this doesn’t happen. Anyone care to bet? Coluccio: I’d like to clarify a point that I made earlier when I agreed with David’s assertion that the PSTN POTS services and wireless/cellular network services should not be lumped together. I initially agreed, but for reasons that are probably no longer valid, and very likely looking at the problem from a different perspective. Today cellular carriers, like their landline telco cousins, share the SS7 architecture. This is unlike the way in which they once communicated their signaling requirements between resources in the past, through IS-41 X.25 approaches, which were not compatible with SS7. The larger wireless operators, in fact, are, and have been, deploying their own SS7 nets in order to remove their dependence on traditional Bell Company signaling, although today most of the dominant wireless carriers are actually owned by the larger Bells, anyway. Reed: SS7 may be used by cellular carriers. But I was talking about handset services. Another example of confusing the carriers with the end-user functionality. COOK Report: Francois in a voice conversation at least three weeks ago mentioned to me his belief that ENUM implementation would somehow replace SS7. (Doubtless I did not adequately understand what he was trying to convey.) Shockey: This is the public ENUM (AKA RFC3761 in the DNS) vs. private infrastructure Telephone Number to URI translations (the so called Carrier ENUM using SIP or whatever). All PSTN calls require some form of number translation to route. The issue is what is the database mechanism to perform those translations and what is the query method if you need it in an IP endto-end world? Of course that assumes (a) you believe phone numbers are useful and (b) some form of interoperability with the PSTN is necessary for some indefinable amount of time. The COOK Report on Internet May June 2005 Reed: To the extent that your blinders only lets you see the world from the perspective of “carriers” (and now I’m including wireless and fiber, to include Francois’ blinders, which are typical of the optics-heads) you miss the essence of what is happening in the world. Despite the enormous “narcissism of the operators” (who cannot see that the world of communications is NOT about them), the real world of communications is about the messages, not the bits, and the users, not those who would claim to own them. COOK Report: We had a similar argument about 6 weeks ago. Is there any chance that we could agree that - like it or not - for the foreseeable future (several years?) SOMEONE besides just the end users will have to operate networks and provide wide area service? Eventually mesh nets and viral networks may combine with dynamic optical technology to push standard operators into the so-called dustbin of history. But for some number of years this is not likely. In my opinion at least. A Changing Role for Operators Reed: Of course operators will exist and operate networks. That isn’t my point about the “narcissism of the operators.” The narcissism is the idea that they deserve to play the role of defining what the services will be that are supported by the meaning-free bits that traverse their networks, among interoperable endpoints. Francois is committed to putting in pipes to carry those bits. I suspect he would like his pipes to be the preferred solution for as many bits in as many circumstances as possible. More power to him, but my point is that other than doing a good job on those bits, he has no leverage to force certain kinds of bits to traverse his pipes. Bits that make up video streams may travel over his pipes or over wireless connections or via satellites or over packets delivered using Bit Torrent asynchronously. The users now have the leverage created by interoperability, and they are using it. They do “route around” attempts to tie services to particular sets of pipes. There’s a fantasy out there that fiber is so wonderful that the first fiber to a home will “own” that home forever, and allow lock in. I don’t buy that - it’s just another way of saying that no one will ever need more than 300 baud because a person can’t read that fast. Shockey: Well if you put 100 channels of High Definition NFL football on it you just might. :-) Reed: If the operators get over their narcissism that the world depends on them, so they should be given all kinds of rights and privileges, they have perfectly fine businesses to run. I’m sick and tired of the wailing and crying and poor mouthing I hear from operators. The sky isn’t falling. Shockey: Well, the sky is falling if you are a LEC and look at their gross revenues and margins on the landline sides of their business. Its not a pretty sight when you then factor into your business model that you will lose 1/3 of your residential business to Cable Operators within 3-4 years and perhaps 30% of your enterprise business to SIP trunking at the edge and IP Centrex from the IXC’s -- if they can hold on long enough. Reed: It’s not the operators that will define the services, or “permit” them to exist. They will get paid, to the extent that a lack of capacity will find entrepreneurs willing to fill it. It’s just that the current operators need to realize that they haven’t got a lock on that new business, and to get started on figuring out how to remain competitive. Shockey: David, it’s just that traditional telephony carriers do not want to be turned into gas, water, or electricity utilities where they have to live on reduced but highly regulated rates of return much lower than they are accustomed to though I need to research this more. The ILECS still think they are high tech businesses when they are not... they are bit pushers. They are totally enamored of “new 39 services” or “content delivery” where we both know that those services, as Anrew Odlyzko has proven, are not all that profitable and you know can be delivered at the edge by anyone. Reed: Else they are going to be the next Kodak or Polaroid - companies who thought the film stock and patents they held were the core of what photography was going to mean to its users forever. Now it’s about ink, image sensors, metadata, coding, and digital storage. Shockey: Well, I still think there is a good business being a bit pusher. The ultimate economic question I have is since we can now conclusively prove that Voice is simply a edge application on the network (the Skype demonstration), which means its real marginal value is zero (like email). What will happen to the telecom industry when you yank over 200B in revenue right out from under it? OK ..50 Billion. Have you looked at the bond ratings on Qwest recently? Coluccio: David, I’ve tried for decades to remove what was admittedly a Bell shaped helmet hairline from my head, to the point where I’ve just about gotten rid of it today. Just enough of it remains - no matter if it’s a common carrier or a COIN - for me to know its importance in tying together communities of users and their endpoints. I don’t think that I am blinded nearly as much by the carrier component of the equation as you have stated. Reed: Frank - I am assured by my Motorola colleague Phil Fleming that pushto-talk in the cellular world doesn’t use SS7 - it uses IP. And of course the 3G world is not terribly focused on SS7, either, since it is moving to IPv6. Shockey: Correct. Right now the classic push to talk application used by Motorola with Nextel (their principal wireless customer) is somewhat proprietary but is essentially an IP “presence based” service... most of the carriers including Nextel/Sprint, etc., are moving to SIPbased presence to enable click to talk COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA across carrier/domain boundaries. happening is a direct reflection of the expression. desire for first of all - carriers to deploy Reed: I wouldn’t bet that cellular end- more profitable services. (By the way I think it’s interesting to mention in passpoint services will spend a lot of time Nextel has the highest ARPU of any ing another clashing set of principles, trying to deal with upgrading of SS7 to wireless carrier based solely on the C2T since the FCC has identified PTT as a encompass new services like SMS, pres- application.) There is also a carrier de- telecommunications service, and as such ence, location-based services, etc. sire to kill off SS7 due to its high cost subjects it to a class of scrutiny along the both the equipment side and services lines of CALEA and other government Shockey: You are correct again. No, side. Compare the prices a Tekelec SCP monitoring and “wiretapping” considthey won’t. They will convert their entire vs. a SUN box and you will know ex- erations. Do you or others have any internal 3G signaling architecture to SIP actly what I mean. thoughts about this? itself. In fact most of the landline VoIP carriers are looking at similar moves Coluccio: Granted, David, the migration Shockey: On this issue I’m pretty cereven for things like LNP CNAME and to IP for all forms of both wireless and tain the Government will have complete 1800 data. SIP itself is a perfectly good landline appears inevitable and on track, access to any form of communications query response mechanism for this pur- but as long as there exists on any session steam they want from any type of service pose and I actually believe SIP will re- a need for one handset or appliance to be provider provided they go through the place SS7 in the core of the NGN. connected to a public wireless realm and normal warrant process. The new trigger the other to a class five on the PSTN, date I’m told by knowledgeable observReed: Yeah, SS7 will probably be around we’ll see SS7 somewhere in the middle ers is the sunset date for the Patriot for a long time in vestigial form, just like doing look aheads and data base dips, Act which I’m told will let the LEA’s, human vermiform appendices. But my probably through the support of dual through the Patriot renewal legislation, point was that rapidly moving markets mode signaling, for some time to come correct the stupid mistake they made in and innovative, high margin applications as you note. It’s my understanding that CALEA by not including information don’t wait for legacy technology tied to most forms of PTT also currently use services authorization. obsolescent business models to keep up. IP in the main here in the States, while They deal with them pragmatically. some are still dependent on SS7 in some And frankly I don’t have a problem with international areas, but those too will this. CALEA was very very poorly writShockey: And cost-effectively. What is make the “switch,” if you’ll excuse the ten legislation. 40 Symposium Discussion January 5 to 15 2005 Some Broader Issues Involving the Blurring of Boundaries of VoIP in the Telephony, Wireless, and Enterprise Worlds Highlights A Balkanization of Personal and Enterprise Communication Trends Coluccio: The discussion that has been about losing the telephone number in favor of a Skype tag or other “personalized ID” I find to be somewhat of a distraction, while also a bit amusing, because it once again demonstrates how the discussion on this list has a tendency to balkanize the telecom universe, and ultimately focusing primarily on “personal” communications trends, as opposed to how those trends will eventually meld with the communications traits exhibited by large commercial and government service characteristics. Yet, wide-sweeping projections are made that appear to encompass all of the above, when they are, in fact true, but they are true most likely for only the most personal level of individual users’ communications needs and provisioning, even if they find their way into enterprises. Like a PC is personal to each user, e.g. Skype-like applications, I firmly believe, will come to permeate every aspect of personal and office communications, but those applications will only “displace” the traditional attributes of earlier services in a limited percentage of total point solutions. Where the newer Skype-likes will dominate most demonstrably, in my opinion, will be as incremental personal applications. Just as email and IM have grown to become, today. That is a far cry, however, from putting telephone numbers in their graves. The world as we know it is not made up of IP coders and IETF delegates, despite the size of this list becoming what it has, and despite our losing focus of the larger flows that take place each day that have absolutely nothing to do with the open Internet, and never get counted on anyone’s stats because they take place behind closed optics. There are folks of all persuasions out there, in all walks of life and business who will welcome Skype and its peers with open arms, and they will do presence and IM until the cows come home, but they will still have a number to call their own at the end of the day,.Either on a wireless unit in their pockets, or on their kitchen walls. And if not at home, then in their offices or places of business. I’m reminded of other times of transition, as when WU Teletypewriter Services Called TWX were retired in the Seventies, and folks wanted to know if Fax messaging would ever carry the same weight as a TWX (pronounced like the candy bar, for you youngsters out there). Later, when email arrived in a commercial context, the same was asked about it, comparing its legitimacy to that of a fax, with the latter owing its credence to a time stamp at the top of the page accompanied by an originating “telephone number.” Of course, my crystal ball only extends out by about five to ten years in areas such as these, so don’t hold me to what I’ve stated above when Verizon is just about finished wiring up Pennsylvania for broadband in the year 2015 ;-) Earlier Coluccio: Some great observations, David, and I obviously agree with you to a certain distance. I’ve often made comment here about how discussions seem to unwittingly divide user populations with respect to the writer’s point of reference, sometimes using generalities to describe what appears to be so obvious, but at the same time often ignoring other aspects of the discipline. In your last message, to which I here reply, you referred to “personal” applications as opposed to large business uses of the PSTN and wireless services, which you so correctly differentiated at the semantic level. As such, many of the things you stated are, or shortly will 41 be, true, where they aren’t already for individuals. But how do those same anachronismworking factors play to the commercial world, I wonder, where NXX numbers are held sacred to large institutions to the point of threatening their providers to change carriers if they were to be changed? I see this as a further widening of the worlds of personal and commercial telecoms, over time, where what appear to be generally applicable statements in one sector simply don’t apply, or much less so, in another sector. Matson: David - Good comment! It has always fascinated me how the world has managed since the beginning of time with individuals choosing their own “ID” - i.e. their name... and yet the telecoms operators and the regulators still cling to a command-economy attitude to allocating identity numbers. One of the coolest things about SKYPE for me is that it permits me to chose my own name/identity in the virtual world just as I do in the real world... no SIP service does it quite like that does it? Forster: Well, you get to choose your own personal name, but the choices you get for business, DNS, and email names are significantly restricted, and these days increasingly involve trademark assertions and lawyers. I imagine the same thing will happen with most any widely used ID, including Skype names. Matson: I was ONLY commenting with respect to individuals, for when you get to corporations, there is the legal requirement (at least in the UK) for a wholly unique legal name - to prevent “passing off”. But coming back to the individual’s self-chosen identity, there are thousands of “John Smith”s in the world or “Mary Brown” - and yet we cope very well in being able to identify and communicate with the one particular one we want - de- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA spite them having chosen a non-unique label. And I see the same happening with SKYPE. Often by the addition of a number (e.g. “Johnsmith445”). It is this autonomy to be known as “I want” - with no subservience in terms of identity to the network, which is (at least to a mere mortal like me) quite appealing about SKYPE. It’s not a big issue - but I do believe it adds just a little to the “delightful end-user experience.” Forster: Also, many times by paying extra you can choose your own phone number? So maybe the two situations aren’t completely different, after making allowances for the 50+ years difference when the systems were designed? Menard: For all I care, my ISP is shipping G.711 audio over my cable modem connection on my VoIP asterisk-based trial. I for one believe that there is much better audio quality than pesky POTS to do with 64 kbps of information. When I am saying that the real battle is open access, I am being serious. You will see innovation coming from those who are not barred from innovating, and that unfortunately excludes most US ISPs under the current Powell FCC administration. I think that we are not putting the critics at the right place. The FCC is killing innovation, not driving it. to this today would have both a technical and business model component to it. Coluccio: I can’t speak for Skype, but I can tell you that during the time frame 96-97 I was a founder in a startup ITSP and faced some similar considerations. As relates to the matter of voice quality, we, too, considered taking the route of a stand-alone algorithm in order to achieve some of the benefits of wider frequency response, fidelity and robustness, such as Skype has done, despite there not being anything at the time that availed itself to these ends. There are obvious tradeoffs to be considered when deciding to use more of the real estate in silicon on quality versus legacy compliance, because once you decide that you’re no longer going to meet the round peg in the round hole requirement - which is tantamount to meeting DS-0, or ADPCM-like form factors and time periods - then you stand to be viewed as an ogre and ostracized as such, as it were, from the rest of the connectivity- oriented universe. That is, until the ‘Net catches up with you and opens itself up by losing its dependency on the PSTN’s form factors and other nuances, which follows the path, apparently, of what Skype has done. And more power to them, I say. Multiple VoIP Markets On December 30 Schulzrinne: Having read some of the discussion, it seems we’re simultaneously talking about three different things: Earlier Reed: If SIP is a good technology, then I would ask why Skype found it necessary to use something else. - product design Jennings: I did ask. This was before Skype had PSTN connectivity stuff. The answer I got at the time was they just needed something very simple and did not need all the features of SIP. They perceived much of the complexity of SIP to exist because of the PSTN connectivity. This was not under NDA and came from an engineer there - not one of their executives. - protocol design I doubt their answer would be the same today. I would guess that any real answer - wideband codecs Only the last one has anything to do with SIP. It is pretty clear that many SIP implementations have fallen short on the configuration end, with lots of configuration options that shouldn’t be necessary to be exposed at all to normal users and inconsistent labeling of the two items that are really necessary. A good SIP device should need exactly the user’s email address/phone number (for corporate) or carrier SIP URI (for FWD and the like) 42 and a password. Making software look like cell phones or office phones is part of the same User Interface disease that makes certain vendors, even ones with a tradition of UI excellence, convert their media players into jukeboxes and DVD playback software into silvery boxes with 7-segment green play time counters. The whole notion of “skinning” is beyond me; strangely, Microsoft’s Windows Messenger SIP tool seems to have largely escaped it. As far as I know, it predates Skype. I guess early automobile designers couldn’t resist making their horseless carriages look like, well, horseless carriages. I should point out that the Pingtel phone, a SIP phone device, was designed with wideband codecs in mind, although I’m not sure how widely they got out of beta with that feature. Jennings: Pingtel had all the same GIPS codec’s as Skype - and the same voice quality. Part of the voice quality is from the codec but another part is from the packet concealment, the gain control around microphones and headsets, detecting if there is a noise canceling microphone or not, and the echo cancellation. Schulzrinne: We clearly have at least two different VoIP markets: first-line landline replacement (Vonage, AT&T and the like) and the talk-cheap-internationally-by-PC market. User feature requirements for the former seem rather different than for the latter. Nigel Ballard: I’ve only been following this thread with one eye so I apologize if this has been covered already. A lot of SIP solutions use the STUN protocol for NAT traversal. That is fine and dandy in the home and SME (Small to Medium Enterprise) environments where at best, the firewall (if there is one) is inbound only. My experience is that STUN will not work through corporate firewalls that are synchronous (inbound and outbound). What I learn from this is that STUN isn’t The COOK Report on Internet May June 2005 a solution for corporate environments, but in the home it is really clever at working through NAT, double NAT and funky router situations on the fly. Reed: Let me throw something into the mix to clarify a crucial technical distinction between STUN and what Skype does. STUN is an amazing kludge (read the spec to look at the definitions of “cone” and full-cone and half-cone and all kinds of ex post, ad hoc classifications of NAT weirdness). It even works, much of the time (but when it doesn’t work, try explaining why to my non-technical friends who just want a simple service). My Vonage phone works fine at the moment using STUN. But the real problem with STUN is that it requires a server. This leads to more complex configuration, and worse, the concept that one needs a SIP ““operator,” even if the operator only provides rendezvous for clients behind NATs that use STUN. What’s neat about Skype is that it is completely decentralized. It presumes scale, and exploits it, rather than trying to solve the problem of one lonely person behind a NAT. Instead of needing a dedicated server, all clients are capable of acting as servers. By distributing the load to the clients, Skype eliminates an unnecessary point of control, and at the same time eliminates a cost of doing business. They can offer a completely scalable service which costs each client no more than they were paying to their ISP already (that’s what “free” means for Skype). This is the same thing Bit Torrent does to eliminate the need for centralized server-farms that must be designed based on accurate prediction of the user load. Now this may sound “bad” to companies used to extracting rents from captive customers by creating central bottlenecks. Skype does make money, elsewhere. But it gives basic end-to-end telephony back to the users, since there is no way to preserve any margin there, now that IP addresses provide universal connectivity. There’s no need to sell oxygen either - the atmosphere is an “Internet” of oxy- gen supply, when we cooperate symbiotically with plants. Regarding Henning’s comments about confusing issues: SIP doesn’t stand alone - its success or failure depends on a technical and business context that define its suitability to task. It’s just not sensible to modularize the world by saying “my part works just fine - it’s the fault of the other guy that my genius doesn’t show through.” Designers need to connect all the dots - that’s what makes someone an engineer rather than a mere technician. Skype Interconnectivity January 5, David Reed: Well, things are moving fast, I just got this: NEWS RELEASE "pulver.Communicator" now Connecting to Skype pulver.Communicator combines instant messaging, VoIP, video, presence, and social networking, and also enables multi-party IM chat across multiple IM networks Consumer Electronics Show, Las Vegas, NV, January 6, 2005 FWD Communications announced today that pulver.Communicator(TM) (www.pulver.com/communicator) an innovative new software product that combines the best elements of Internet-based Communications (instant messaging, voice, video, presence, and social networking) in one application now interoperates with Skype(TM), the popular VoIP service provider. With pulver.Communicator beta version .94.3, users can import Skype contacts into the application and conduct voice calling and instant messaging with those contacts, as well as engage in multi-party IM chats amongst friends across multiple IM networks. “I’ve been a big fan of Skype since its launch and decided to create a link between the two applications”, said Jeff Pulver, chairman of FWD Communications. “pulver.Communicator is the first outside application to offer Skype functionality to users of the product, which is quite significant as it opens the door to a multitude of friendships and relation43 ships across the most widely-used Internet networks , (AOL, ICQ, MSN, Skype, and Yahoo!).” Communicator Features Users of pulver.Communicator can communicate via voice, video, or text Instant Messaging. Specifically, the application enables users to: Make free calls to millions of VoIP users worldwide (not only to other pulver.Communicator users, but also to subscribers from the more than 80 VoIP service providers with whom FWD Communications has established peering partnering agreements), as well as to a wide variety of Free-Call services in the US, UK, Holland, Japan, France and numerous other locations Initiate onthe-fly conference calls Establish personto-person IM sessions with other pulver. Communicator contacts, as well as between contacts on the four most popular IM networks (AOL, Yahoo!, MSN, and ICQ!) Establish multi-party IM sessions that cross IM service provider boundaries Place video calls Enjoy and expand their personal networking. Through the Shared Buddy List feature users can view the buddy lists of other subscribers, and they can share their own (of course, pulver. Communicator can be easily configured to limit or completely restrict contact information sharing at the user’s discretion). Personnel from pulver.com will be conducting demonstrations of pulver.Communicator and its many features at CES, including the new Skype interoperability inside the pulver.com sponsored VoIP Tech Zone on the CES show floor. Since its official launch in October pulver. Communicator has been downloaded more than 50,000 times. Getting Started pulver.Communicator is free, and it can be easily installed on any PC running Windows XP. To download the program, please visit www.pulver.com/communicator. pulver.Communicator is also available in Spanish. Coluccio: David, you brought up a few points, probably inadvertently - but I’ll take what I can get when an idea serves me well - in the following segments of one of your last messages: COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA "There’s a fantasy out there that fiber is so wonderful that the first fiber to a home will 'own' that home forever, and allow lock in. I don’t buy that - it’s just another way of saying that no one will ever need more than 300 baud because a person can’t read that fast. [...] "Now it’s about ink, image sensors, metadata, coding, and digital storage." Meshed Up Neighborhoods Several times I’ve broached this point, which seems never to gain any lasting traction here, so I’ll try it again. Imagine cell sites that don’t always talk back to base stations, but, instead, overlap and pass traffic horizontally to one another when the logical flow of traffic dictates. Now imagine an identical scenario, only this time it’s fiber based, designed to achieve the same osmotic type of flow between neighboring meshes made of glass, or made to hybrid with wireless, as well. Through extension, such a model would allow crawling to take place, albeit at light speeds, in a space previously thought of as the edge of service providers’ access networks, thus creating and facilitating a new core for transport between non-adjoining communities of any given type, to include industrial parks, schools and campuses, bedroom communities, or wherever. Such a model could never even be contemplated through the use of a first fiber to the home scenario such as you’ve mentioned, because such a model dictates its own limited set of provider-intended services, as you’ve noted. Hence, never supporting the group communications modes you’ve articulated, either, without first going back to head ends and central offices, or base stations, first, to “check in” and “check out.” I submit that a passive form of media extension is possible allowing for a conjoining of meshes, within limitations imposed by distances at first (and then even where distances are excessive in future generations, through gang amplification techniques), but will never see the light of day unless brought to a public forum for examination and trialing. While I’m not about to carry the torch for such an undertaking at this time. I do spend some time wondering, however, how you and other list members here would view the prospects of an architecture such as the one I’ve outlined, above. Noting that what I’ve suggested here doesn’t fall very far, in principle, from where the adhoc wireless topologies that have been proposed and are now trialing at this time are headed, under the umbrellas of Wi-Fi-lookalikes and later iterations of WiMAX. In considering such a radical shift in architecture for the more traditional forms of transmissions, such as xMDS and fiber-based access nets, keep in mind of course, that all first- and last- mile vendors, several of whom are even represented here, have based their entire design strategies to support the service providers’ current modus operandi, and that endeavoring in the open fields with either cell-based wireless sites or fiber optic builds is no trivial feat to attempt without doing ample levels of due diligence, first, and a firm resolve to see it completed. Reed: Cool - perhaps some of these ideas might be worth checking out! I’d only add that in some cases, at least, an entrepreneur can do this sort of thing incrementally, not having to deploy it universally. And the mindset of risk capital lets it place bets where the payoff need not have 100% certainty. The Achilles Heel of a Verizon is that it can’t catch all of the termites tearing up its foundation (especially when there are a wide diversity of termites, each trying something different). As long as its brains are concentrated in an out-oftouch management that travels in private jets, rather than pulling cable or using a spectrum analyzer, it will have the same view of its demise that AT&T Bell Labs management had of the Internet and its architects (a cute, irrelevant toy built by people with bad grooming habits who wouldn’t be allowed to play golf in Short 44 Hills, NJ). Coluccio: I should have prefaced my earlier “meshed up neighborhoods” message, lest it otherwise appear to be a harebrained notion, by stating that I am fully conversant with conventional, tried and true, link level techniques used in loop access systems in DOCSIS, FSAN, 802.3ah, and some of the free form variants, thereof. I use this disclamier in order to discourage those unfamiliar with me from chalking my hypopthesis up as mere noise coming out of left field, which it is not. Whether anyone would ever mobilize and attempt such an undertaking is another question, due to the perceived risks involved as David noted, and the as yet unarticulated parameters required to make it work. David, we’re in agreement in that any attempt at achieving the goals that I’ve suggested should be done only after extensive modeling, and even then, only incrementally, as you’ve suggested. I’ll continue to give this model of interlocking, honeycombed nets comprised of tiled polygons some further thought, perhaps putting it into a more coherent workup and revisit it here at some future time. In the meantime, if anyone has any thoughts they’d like to share on this subject, please feel free to post to them here, or contact me off list. TIA. Ballard: I recently consulted on a project that involves deploying commercial Mesh today as a stop gap technology for FTTD in two years. On the surface it would seem that fiber into the premises is enough of a deterrent and has enough available bandwidth to ward off all other technologies. The practice I think is somewhat different. Said fiber is typically controlled by someone, so any services that ride on it are subject to fees and possibly other non-competing restrictions. Other delivery methods, typically a wireless CPE (WiMAX or otherwise) on the premises gives a competing path into the property albeit with hugely reduced bandwidth options. But the infrastructure required is minimal in comparison. The COOK Report on Internet May June 2005 I guess it all boils down to how good the fiber company is at providing enough compelling and competitive services to keep everyone else out of the picture. Coluccio: Nigel, you wrote: “I recently consulted on a project that involves deploying commercial Mesh today as a stop gap technology for FTTD in two years.” I assume that the D in FTTD doesn’t stand for “desk.” What does it stand for? It’s a bit uncanny that you should mention “in two years.” Do you mean by this that you ‘pre-wired a mesh network today in advance for its expected use two years out? Please explain. I say it’s uncanny that you mentioned two years out for this reason: Since the state of the art of optical component technology is not yet at the stage where it would need to be in order to fully support the price performance criteria making my hypothesized topology feasible, I too have thought of a planning stratagem whereby a trajectory of convergence between the physical media installation and networking capabilities at field nodes would coalesce in about two years, or thereabouts. You stated: “I guess it all boils down to how good the fiber company is at providing enough compelling and competitive services to keep everyone else out of the picture.” Interesting point, because it brings to the fore one of the considerations that we’ve discussed here numerous times on this list, most recently with the help of Tom Hertz of Fiber Utilities of Iowa (FUoI). (I’ve copied Tom’s introductory message to this list at the very bottom of this page, following Nigel’s script.) And the basic problem that raises its head immediately, of course, and this isn’t new news to anyone here, is: Who should decide what the physical L1 through the routing L3 attributes should be for any given locale or region. Left to incumbents, we already see what we get: Lots of hub offices supporting star topologies working off tree and branch topographies, all of which - whether through design or through happenstance - assuring that the toll booths keep ca’chinging. Interestingly, in large part many of the dollars “earned,” not to mention the latency that is introduced, at those toll booths go to paying for the trips that packets take to get there, and then very often back to other users on the same network or to users in adjoining communities. And I’m not anti-head end or CO when I say this, but rather treating the need for going to those places with a more prudent eye. COs and colos are indeed required for backhauling to the core, for storage and server purposes, but not to the extent that everything that I send out to my neighbors need necessarily traverse them, much less be governed by what the service providers manning them want to allow me to do. Don’t get me wrong. I’m a big believer in physical layer bit pushers getting paid. I just happen to also believe that those who are in the drivers’ seats today could be doing a much better job, or should be planning to do one down the road, than they appear satisfied to be doing today, which, in the main, are practices and designs that are predicated on network attributes that inherit from copper wire network characteristics of a hundred years ago. I’m referring to the distance limitations, phase shift characteristic properties tied to delay (which necessitated loading coils and repeater assemblies every six thousand feet at one point), and the very weight and girth of feeder vs. distribution cables, right down to the neighborhood cross-connect and drop wire. To get a better handle on what I’m referring to here at the bulk physical level, consider the support structures (Hi Francois!) or spaces and pathways required to support a 3,600 copper pair cable that is used to feed a section of a neighborhood, and all of the splice points and punch down cross-connects that they entail, which, today can be replaced by a couple pairs of fibers and some field mounted nodes. 45 Did I hear someone say that there was no room left on poles or in conduits to support the rights of way for independent fiber optic operators? Can a new entrant do anything differently, e.g., laying down a quilt of interlocking meshes such as I’ve described, while remaining solvent in the process of advancing the state of the art in access plant technology? What would that sort of thing take, short of an I2 or NGI level of funding from the feds and possibly a few large, sympathetic industry players looking to get in on such a venture at an early stage? Ballard: FTTD = Fiber To The Door. I’ve noted some folk now use the acronym FTTP (Fiber to the Premises). Yes, their plan, not mine, is to use Wi-Fi Mesh as purely a stop gap solution till the fiber goes live in 24 months. As it stands today, Mesh falls way short in its ability to provide much beyond Internet access and VoIP if you are lucky. But for most people, high-speed Internet is what they crave and mesh can provide it. The speed of provisioning such a system, especially if the city is on board (pole access) can be startlingly quick. It gets city’s where they feel they need to be and fast. And whilst fiber has the technical ability to pump all manner of rich audio visual services into every home with ease, the deployment costs and the related timeline to achieving anything approaching blanket coverage, is high and slow. Fibers high cost and many city’s desire to be placed firmly on the technology roadmap (look a us, we’ve got a killer infrastructure) today has given the mesh vendors a most welcome boost. Make hay boys and girls! Skype Marches On Hughes: As if in confirmation of my belief that VoIP can spread across the Third World faster than e-mail, and help make the “rest of the world” computer literate I made an excellent Skype voice COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA connection with Shiva Kulung, a Rai from eastern Nepal. The Rai are minorities among minorities of that part of the world. Though the famed fighting Ghurkas come from Rai stock. (While the Sherpas appear not to be war-like. I have to find SOME warriors to do in the Maoists) Shiva is a sweet young man I met in Sherpa country Namche, Nepal. Shiva had been hired by Sherpa Tsering as a cook, handyman, and was told to look out for old man (me) while I was there. He brought me hot Milk Tea at 5AM when I couldn’t sleep with the time change, and I would sneak into the cold Cybercafe computer room, off the lodge kitchen, to check and send my email. He could not speak English well at all when I was there, much less read or write it, or use computers. Like many a Sherpa Lodge owner who has made a little western money and has to cook for and serve trekkers and climbers, they hire a lower class than they are - principally Rai’s - whose homeland is in the northwestern corner of Nepal, are largely Hindu, not Buddhist. But now a year later, and especially when both Tsering, who owns the place, and Santosh, his Nepalese Tech have been gone in Kathmandu for almost a month, the whole Cybercafe has had to be operated by young Shiva! Who has learned progressively, was doing MSN Instant Messaging with me every so often but no voice because its sound would not work there. But he now has figured out Skype! And contacted me first by Skype’s IM messaging utility then I found his Skype “‘handle” (shivakulung) in the search utility, with his ”name” as shivarai and called him voice. Worked perfectly, even through the two satellite and wireless 64kbps hops from Namche to Kathmandu and KT to the US, with 1,200 ms latency. He was thrilled to be able to talk with me, clear, understandable voice. Giggling all the while we talked. (The young man may be lonely, he is a Hindu Rai in Buddhist Sherpa country, at 11,500 feet, no family, much less friends, or girl friends, near.) So every evening about 10PM his time (9AM MST here) he tries to chat with somebody ”out there.” Now he can do it with voice. All of which confirms my view that, with sitcom, unlicensed wireless out from the base unit, and Skype, that the Third World will be chattering by voice long before they learn, or get into, “email” which requires a degree of English literacy even more than voice. Shockey: I don’t disagree but lets not forget that Skype is a private vertically integrated application with its own private naming and addressing scheme unique to Skype. I’m sure there are more Skype’s coming down the road which means that we will have the same kind of fragmentation that we saw in IM with four big services unable to talk to each other. SIP was designed as a global service using global naming and addressing (the DNS) etc. Hughes: Next time I chat with him I will ask him WHICH sub tribe of Rai he is. For my research shows me no less than Bantawa Rai; Chamlinge Rai; Danuwar Rai; Kalinge Rai; Kulunge Rai; Saam Rai; Tamachhange Rai; Thulunge Rai! Each with their own dialect of Rai, a little Nepalese, and very little English or other western languages. I wouldn’t be surprised if Shiva is the first Rai ever to chat VoIP over the Internet. Access to Broadband Baller: USA Today Endorses Lafayette FTTH Project http://www.usatoday.com/ news/opinion/2005-01-10-broadbandbellsouth-our_ x.htm Coluccio: Jim, USAToday is certainly getting some mileage with this one. It’s good to see topics like this covered in depth. Thanks for posting the article. I note that the title’s use of the term squashed appeared to be a bit incongruous with the story. And the story even appeared truncated, because I don’t necessarily 46 see anything being squashed. I should also note that today’s edition of the paper reveals a contrasting view by Bell South’s William A. Oliver, president of Louisiana operations, BellSouth Corp., New Orleans. Mr. Oliver’s contention being? What else? He says: “Competition Should Be Fair” http:// makeashorterlink.com/?Z5F32463A Baller: Thanks, Frank. We are including Bill Oliver’s response to the USA Today editorial in our daily email, but I wanted the get the editorial out undiluted first. Attached is Bill Oliver’s testimony to the Louisiana Senate Commerce Committee in support of the compromise legislation that we negotiated last Spring. In particular, contrast Oliver’s whining in today’s piece with his testimony that the legislation “insures that appropriate competitive safeguards are put in place to create a level playing field in these competitive industries.” David Sandel: Thanks for the article too, a good and simple discussion surrounding a complex topic. Thinking about this in other economic terms, airports had to be constructed to support the airline industry, roads and bridges had to be constructed to support the automobile industry. Both of these forms of municipal infrastructure (and many others) had to be developed by local and federal government to meet the economic potential of a given technology. With that being said, broad band infrastructure will also need to be constructed by muni’s to meet the potential of the technology and support the economic development and security of a community, just like roads and airports were built. Roads provided equal access to all auto manufacturers, airports to all airlines. For municipal broad band “infrastructure “ to succeed it must provide for equal access to all service providers. The COOK Report on Internet May June 2005 Patrick Leary: This echoes what we (Alvarion) have been advocating corporately and individually for a long time now. Equitable broadband access is too important to wait for the private sector to ride to the rescue. The analogy I like to use is that access to broadband today is much like access to clean water about a century ago. Today, we are at the hand pump stage of broadband. Chronologically, it was a long time between when folks gathered their water by pail from the nearest fresh source to when folks used a hand pump. During the extended “Pail Period,” people used water only when necessary and the definition of what was necessary was very narrow -- drinking, a little cooking, and maybe the once a month bath (if they otherwise did not bath directly in the rivers and streams or had no access to bath houses). But once that hand pump became available outside the front door, well by goodness, water was so conveniently available that usage grew dramatically and what was deemed necessary grew. And then, it was but a small leap from that day to the perception that one had to have taps in the home in the kitchen...then indoor bathrooms... then hot and cold taps. Today, not having clean, fresh water available on demand anywhere people congregate or live is, quite literally, unthinkable -- it has become as assumed as flicking on a light switch. With broadband being at this pump stage, people are discovering everyday uses for their high speed access and whole markets will be created leveraging its eventual ubiquity. It is only a small leap in time from this point to broadband availability being the next assumed utility. Accordingly, like provision of any utility long beyond the pump and certainly the pale period, that the private sector would assert that it has the right to control availability is a dead, or at least dying, argument. The private sector can play a pivotal and even fundamental role, but to claim ownership is, well, over the ”pail.” I hope it is only a matter of time before some wise city or state’s attorney is able to articulate a winning “promote the general welfare” constitutional argument. January 11 COOK Report: I have added Raj Sharma of NexTone to the list. Raj’s principal interests lie in the harmonization of protocols for the IXCs, ITSPs, ISPs and ILECs, such as SIP and H.323 along with a long list of proprietary ones made by all of the VoIP field’s popular vendors. Reed: Cool. Welcome Raj to the debating society! But to get down to brass tacks, and semantics that show a broken mindset, “harmonization” (in its typical telecom meaning) is not going to be a winning strategy. Harmonization is what a bunch of technocrats, relying on a government enforced oligopoly (i.e. the ITU) thinks it has the power to do. I.e. shades of the 3GPP, or NANP. VoIP is inherently happening at the edges (whether they are PBXes, SIP phones, etc. at the hardware level, and Skype, FWD, AIM, PTT, Blackberry email as the viral integrating technologies). Viral Communications Stastny: To which I fully agree. By the way, David, congratulations to the “Viral” paper. Since it is a draft and quite old, does an updated version exist? Reed: Richard - the viral paper was written for a special issue of the BT Technology Journal (October 2004). That journal is now out, and I commend it to you... http://dl.media.mit.edu/viral/viral.pdf may be the same version you have - I think we should probably replace it with the prettier published version from the BT Journal, now that you point it out and assuming we can get a PDF of that. Editor’s Note: The special issue of the BT Technology Journal is on the Media lab Web site and is well worth reading. However I steadfastly disagree with David in that I find the general sweep of the May 19 2003 draft much better than the two early articles in the BT journal 47 that are intended to replace it. Reed: You may also be interested in our group website: http://dl.media.mit. edu/viral/ and the cross-MIT sponsored program Andy and I have launched with Dave Clark and Charlie Fine called the Communications Futures Program, which focuses on the evolution of the architecture of the communications industry (website under re-construction). And the broader embedding of the viral work in human networks, our bigger vision called “Organic Networks” in that special issue: http://dl.media.mit. edu/BT-vco.pdf Don’t want to be seen as hawking CFP or the Media Lab here, though. COOK Report: David, I’d like to do that for you. :-) Seriously I am amazed - People on this list are gradually beginning to read it and uniformly say WOW. I advertised and endorsed it to a private list of David Isenberg’s that some of you are on there again not much reaction. Last night I sent the PDF to Sebastian Hassenger and pleaded with him to read it. Talked with him today. He has read it and remarked that he finds it really wonderful and wondered how he ever missed it. Sebastian is Senior strategist for pervasive computing in one of IBMs WestchesterCcounty, NY Labs. He was on one of my lists in the August October 2003 time frame. To the extent that I understand what he does it is to survey where all this stuff is going and make sure his colleagues at IBM understand. I told him what we are doing and he agreed it was right down his alleyway and agreed to join. David, there seems to me there is a theme going here... many, many people who know you and your ideas have turned up absolutely unaware of this paper. The ideas are certainly out there. But the May 2003 version of the paper wonderfully ties them together in a FRAMEWORK or a pair of lens through which to see the world. We need frameworks I think. And COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA this is a REALLY REALY GOOD one. Now you mentioned special issue of the BT Technology Journal (October 2004). With that as a lead, I found in Google http://www.media.mit.edu/publications/ bttj/ This URL brings up the table of contents for the October issue, It looks like well over 300 pages and, yep, the whole shebang is downloadable. The first 6 of I think 27 articles are listed thus Organic networks: A Lippman and A Pentland Viral radio: A Lippman and D P Reed Wirefree in Patmos: M Bletsas Sensate media - multimodal electronic skins as dense sensor networks J A Paradiso, J Lifton and M Broxton Collaborative knowledge building by smart sensors V M Bove Jr and J Mallett SimPhony V Lakshmipathy and C Schmandt The listings don’t look like URLs but they ARE. Click on organic Networks and you get the article as a pdf. You can get them all this way. Now David I cannot find the specific May 2003 publication in this BT Journal issue.That looks to me because you seem to have rewritten it. Taken the more general ideas and put them in organic networks paper and taken the radio parts made them a lot more technical and put them into viral radio? I have done only a quick skim but are my conclusions correct?In any case this October issue is a gem so thanks for mentioning it. But in my opinion the may 2003 for people who aren’t REALLY technical might be the better starting point. “Harmonizing” Discordant Systems Reed (earlier): Harmonizing is what the band did on the deck of the Titanic as it sank. The driver of coordination of standards will be voluntary cooperation among market-leading innovators, not backroom deals. The value added by carrier-centered protocols is minimal, and the carriers who think they have the power to “define” and “harmonize” will go the way of ISDN and AIN. At best, “harmonizing” will involve building bridges between independently developed networks. Merely middleware, and as some of my smartest VC friends have told me, middleware is not often a high-return business plan, because there are lots of middles, and “middles” rarely have a sustainable technical or market advantage. So in a high-growth, rapidly changing industry or a rapid train-wreck industry (like voice), don’t plan for any middle role to send your children to college (unless they are already there). Coluccio: Thelonius Monk chord coming off a Charlie Parker solo, wrote: But to get down to brass tacks, and semantics that show a broken mindset, “harmonization” David, there were reports of loud bursts of laughter in my immediate vicinity just a few moments ago. I am guilty as charged of sending those words in a stream of (sub)consciousness to Gordon when I introduced him to Raj. I request that you do not… repeat, DO NOT hold this against either Gordon or Raj, for it is I solely who should be sent to trial for my selection of words. This little exchange has now caused me to seriously rethink the time required to lose my bell-shaped hairline. I thought it was gone, but I guess not. Having said that, it’s difficult sometimes to have to go to work every day and still agree with what you’ve had to say. One of my clients, not so long ago, was five banks and two brokerages covering 182 countries plus the Vatican, and now today as a result of acquisitions and 48 mergers they are one. They suddenly find themselves with quite a few PBXs, routers, telephone sets, core networking philosophies and theological twists, not to mention probably every VoIP protocol and compression algorithm ever written, to manage every day. In a way, they resemble a paradigm of yore that one can find painted in your note, except that, from a networking persepective, their network is now larger than many of the telcos around the planet that make up the ITU. I’ll send them a snippet of your note telling them how arcane their network strategy is, and then I’ll pass along what they have say to me, in return ;) Finding myself feeling suddenly alone, where before I was a tenor in a four-part acapella group sitting between a rock and a hard place, Dave O’Leary: Sounds like the discord is between the optimism of where (some of us) want to be and the pragmatism of where (many of us) actually are. (Using “us” in a broad sense, including all stakeholders in the networking/ telecom community, not just the folks on this list). Reed: Frank - I think the metaphor is apt, even for your example. It took a long time for the Titanic to sink ... long enough for a lot of harmonization, and maybe even a fair bit of comfort to those who were confronted by the inevitable. Of course the irony was that the wireless operators were too focused on transmitting the ordinary traffic to serve what could have been a role in preventing the disaster. What does that mean to the harmonizers? Intel Backs Municipal Broadband Coluccio: Intel to back broadband role for cities . By Richard Shim http://news. com.com/Intel+to+back+broadband+rol e+for+cities/2100-1034_3-5532714.html . Jan 11 17:47:00 PST 2005 . “Chip giant Intel on Wednesday plans to provide a high-level perspective on The COOK Report on Internet May June 2005 the ongoing debate over the role of the public and private sectors in providing broadband services. In a speech at the Wireless Communications Association in San Jose, Calif., Intel Executive Vice President Sean Maloney is expected to encourage commercial service providers and public agencies such as city governments and municipalities to work together in building out new broadband infrastructure. Intel has a keen interest in the proliferation of wireless broadband technology and industries using it; by early next year it plans to produce WiMAX chips for networking equipment that carriers can use to sell high-speed Internet access to consumers. WiMAX is a promising wireless broadband technology allowing data to be wirelessly transmitted across several miles at transfer rates of several megabits per second. “Sole responsibility, either from government or a single carrier, of a city’s wireless network is not the best solution for growing the market,” said a source familiar with the chipmaker’s position in wireless broadband policy. “A sharing of responsibilities is what will encourage broadband adoption, and that will be a key point in Intel’s policy proposal.” . Maloney will outline the company’s high-level policy position and will speak out against efforts to ban municipally owned networks. In recent years, phone companies and cable providers have actively lobbied local and state governments to ban public agencies and municipalities from building their own communications networks. The commercial providers have been successful in some regions of the country. In some instances, commercial providers will be able to build networks and offer the best network choice to customers at affordable prices. But in other instances, such as low-income areas or rural locations, it might make more sense for a city or some other municipality to build the infrastructure. “We welcome Intel’s position and strongly support collaboration between the public and private sectors,” said Jim Baller, a principal attorney for the Baller Herbst Law Group and a leading expert on municipally owned networks. Intel’s position is partly in response to strong lobbying by Verizon Communications that helped lead to the passage of a law in Pennsylvania that prohibits cities from offering Internet access to their residents for a fee. Verizon and other incumbent phone companies had urged legislators to ban municipally owned networks to prevent other cities from following the lead of Kutztown, a small college town near Allentown [PA] that set up its own telephone, Internet and TV system in 2002. Phone companies and cable providers argue that municipalities that build and own their communications networks have an unfair advantage because they are backed by public funds. They claim that the municipalities will drive them out of business by offering services at greatly reduced prices. On the other side, communities that want to build their own networks argue that they want broadband services now, and they are not willing to wait until it becomes economically feasible for commercial providers to build the infrastructure. Sandel: Thanks for passing along the Intel announcement. This is certainly a strong ray of light at the end of the tunnel. Hopefully more vendors (and other entities) will begin to jump on board with this message as it relates to the growth of broadband networks. There are several items in the press release that I would like to highlight: 1). “Sole responsibility, ----- is not the best solution for growing the market.” All I can say is AMEN to that! Many service providers feel that muni infrastructure will cut them out of the market when actually it is the opposite. When muni infrastructure (open access) is in place and local regional exchange is possible with other networks a QoS problem is solved that will allow for a very significant growth in metropolitan bandwidth as higher bandwidth services now are possible and desirable. I say open access muni infrastructure will help to drive bandwidth demand for all service providers something they cannot do on 49 their own. This is not competition, it is broad band economic development for all comers. 2). “Communities are not willing to wait.” We are now in an IT industrial age, the communities that have the best broadband infrastructure and services will be the winners when it comes to economic development. 150 years during the railroad age communities FOUGHT to have railroad access.Tthis is no different. This requires municipal support and private cooperation without which the railroads would never have grown to the extent that they were able to. 3). “…strongly support collaboration between the public and private sectors.”. This model has been repeated over and over. Look at the electrical utility industry, the port authority, the international airport, the highway system ---did the private sector pick this up. Did the development of these infrastructures ultimately HELP or HURT private industry? I am excited about Intel announcement (and others that may follow) because I think the scales are on the verge of tipping in the right direction, something we all need! Ballard: We recently brought Intel Consulting in at some cost to assist us here in Portland, Oregon with our municipal wireless plans. While their support isn’t a surprise, it is really nice to see Intel say it publicly. They must have thought long and hard before making such a statement. COOK Report: Indeed, Intel sells a LOT to gear to the phone companies and informally people working there have told me privately that they have had hell to pay internally from Intel divisions that rely on telco sales for advocating the kinds of things that we talk about on this list. This could be therefore an interesting signal. The Window of Opportunity for ENUM is Closing Fast Frode Greissen to Interesting People COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA on January 12 : Last March we got the idea to set up the name server for the Danish ENUM domain, i. e. using Danish phone numbers as domain names. The phone number +45 12345678 would be transformed into the name 8.7.6.5.4.3.2.1.5.4.e164.arpa and then the user could use DNS to direct communication to his IP phone, email or PDA. We were not sure it would catch on since there are so many other domain names, but I work at a government agency whose mission includes piloting new technology, and running the top level would be trivial as long as the volume is small. There is an agreement between the ITU and the IESG that the RIPE NCC runs the top level domain but requests to operate county domains must go to ITU-T which forwards the request to the country telecom regulator for decision. Now, Denmark is a small country with 5 M inhabitants and a well-funded and well-working public administration so at first I called the regulator and we met a few weeks later. They were interested and had questions. As talks and correspondence grew and took time without progress we sent the request to RIPE NCC and a bit later the regulator got the official request from ITU-T. Their response to this difficult situation was to hand over the case to their resort ministry (for Science, Technology and Development). Well, after an exchange of letters we had a meeting with the ministry who explained that they had to make a public hearing. I provided paperwork for the hearing which was sent to 160 organizations. The responses were mostly negative with good arguments that the Name Service should be allocated based on open tender (which would take another year and perhaps attract no bidders) and that issues with security, competition and rights should be thoroughly discussed and understood (which would require some careful work that nobody may be likely to invest on such a risky project). Of course, the organizations who had the time to respond to the hearing include the incumbent telco and other large telcos who have to argue against all change that may threaten their existing business, plus trade organizations who have learned that the Internet got big and that there is money in domain names. The small IP telephony companies did not have time to respond. Right now, I’m trying to convince the ministry that they should follow up with a public verbal hearing and get a decision before the anniversary of the case. In the mean time Skype has been downloaded 50 million times. IP telephony companies will be routed without ENUM, and the ITU and governments should not manage technology innovation. COOK Report: Something though sure seems to be holding ENUM back - opinions? Shockey: Politics - pure and simple. Once the decision was made that the various zones of e164.arpa were to perfectly match the administrative policies and procedures of the ITU E.164 plan the decision to implement became a nation state issue. This should not surprise anyone. Phone numbers have always been administered by national regulators and regulators have to do their job to make sure that various national laws and interests are protected. Decisions of this magnitude take time and careful consideration hence the need for public consultations and hearings. And of course there are some entities who are not happy about VoIP. Richard Stastny: First, I have to apologize for not participating actively in the discussion here on Skype. I was on vacation for three weeks and back on January 10th. So I am still catching up the tons of e-mails. So I was only listening randomly. But with this new thread I am questioned directly, so I have to answer ;-) Frode is perfectly right with his analysis. It takes (too much) time to get ENUM implemented in e164.arpa nationally. 50 The story in Austria is not so much different, we only started quite early, the main reason of this was my involvement in IETF, ETSI and ITU-T SG2. So I launched an internal ENUM Task Force within Telekom Austria dealing with this issue and also presented ENUM nationally. Therefore our regulator made the first round of questionnaires in August 2001! and a workshop in February 2002. There was only one major difference to most other countries: all other approaches first tried to solve all technical and especially all administrative problems before they started a trial - which had the result that they never finished this phase. In Austria we did not forget these problems, but we started the trial in parallel. One reason was that the investments are ridiculously low - all you basically need are some PCs (or some Dell servers, if you want to be exquisite) running Linux and DNS, and a handful of good developers. This is of course not true for the commercial implementation, but it was sufficient for two years of trial. During the trial we found out the following things: -There is no technical problem in implementing ENUM, since it works fine, and is fast and reliable. There is mainly a commercial problem: ENUM is primarily for interconnecting companies, as Richard Shockey found out already two years ago, secondly it is for ENUM-enabled (ENUM-only) nomadic numbers (you need again the regulators coop here), and then eventually for mobile numbers, For all these applications validation (which is used mainly as an argument to block ENUM) is not an issue: not a problem for companies, a non-issue for ENUM-only numbers and easily done with SMS with mobile numbers. Residential customers on fixed lines will come later, if ever. The basic problems with ENUM in Austria are twofold: One is Metcalfe’s Law, both nationally and internationally: Austria is the only country which has ENUM implemented, so whom to call? So we basically wait The COOK Report on Internet May June 2005 and wait for others (especially the US) to join. The second is: you need a VoIP service giving you a public SIP-URI. There are not many commercially available VoIP services which give you a public SIP-URI, Every Telco and most VoIP providers are implementing in a walled garden. Again the only ones which can bypass this are again the companies. SkypePlus - other VoIP providers do this already (and not only in the US). An Austrian company (Kapsch) has implemented ENUM already. The major feature is that they connected their PBX via a VoIP gateway with the Internet and you may reach any extension with sip:ext@kapsch.net eg sip:3184@kapsch.net. So if you query ENUM for +43508113184, you get sip:3184@kapsch.net. You may try it out at http://193.80.173.26:8080/elookup/ If somebody at Kapsch dials any phone number on his office phone, first an ENUM query is done to find an ENUM entry, and only if no entry is found, the call is placed to the PSTN. Stastny: as I said above, many providers offer numbers for VoIP - have a number in Washington from Libretel. This is a national matter and most countries allow this (some with restrictions) The bottom line is simple: the window of opportunity for ENUM is closing (fast). The other solution proposed to implement ENUM fast without dealing with ITU, national regulators, etc. is to do carrier ENUM in another tree. But this would work globally only if ALL carriers and service providers would agree on one common tree (and also connect to the public Internet), and this seems not possible either. The argument against ENUM: Who needs numbers at all, we use names, is ridiculous: What names: SIP-URI, Skype names, AOL names, H323-URI? The basic problem of ENUM as stated above is that you need to translate the number in a public available URI. If you do not have such an URI, you simply cannot reach the other party, with or without ENUM. Period Roberts: With number portability (at least in the US) would Skype be able to take the numbers? This would remove all barriers to entry. Stastny: This will come very soon with Roberts: I doubt any country will allow this easy a revenue source to shift to other out-of-country companies. And since as Richard Shockey correctly points out, this is the responsibility of the ITU, and they have passed it to the countries, Roberts: Enum is doomed until the VoIP conversion is near complete. Stastny: One does not need ENUM for providing these services Roberts: Then it is too late, we will be using real names. Stastny: You need always an URI as I pointed out in my other e-mail, both for ENUM and direct addressing. Reed: In my opinion, those who think that numbering plans are either a barrier to innovation or an enhancement of social welfare are doomed to live in the dustbin of history. Do you think any normal human being wants to continue to have to manage a list of numbers for telephone instruments designed around the limitations of a rotary pulse generator or a Strowger switch? Of course the idea that interfering in speech communications among people is a key governmental role is popular, even in strongly democratic traditions like the Danish and the Americans (CALEA comes to mind). The more bollixed up in politics that ENUM becomes, the better. Two Parts of VoIP COOK Report: Please welcome Raj Sharma of NexTone. As Franl Coluccio remarked above, Raj’s principal interests lie in the harmonization of protocols for the IXCs, ITSPs, ISPs and ILECs, such as SIP and H.323 along with a long list of 51 proprietary ones made by all of the VoIP field’s popular vendors. http://www.nextone.com/ Menard: To kick-off, perhaps Raj can enlighten us on his contentions of using a SIP-based border proxy server with and without an RTP proxy and what this means for settlements and accountability for inter-carrier IP-based peering on a bill-and-keep-basis on a per-exchange basis. Reed: Francois, with all due respect, this is micro- (no, let’s say pico-) optimization of a system for which none of these concepts may be applicable. I’m sure somebody cares, but I’ll ignore any future email that addresses any of the above, little-picture obsessions that represent the death rattle of a lot of obsolete ideas. (And good riddance, some might say) COOK Report: Hi David, Thank you for emphasizing again that we are talking here about two different strains of IP telephony. Like it or not there *IS* a sizable market of telcos, and enterprises with IP PBXs and so on that are running VoIP systems of one kind or another over the OLD legacy telco infrastructure. At the same time you have SKYPE that seems to represent a new way to do telephony or better to say voice communication and has the characteristics for which you are a brilliant advocate. One reason that I have flogged your Viral Communications paper so heavily is that someone who reads it will understand VERY clearly why you are as outspoken on these issues as you are. However - in MY opinion - if you look at the marketplace for the next couple of years, it is pretty clear that like it or not the pico optimization of these issues is happening and will continue to happen. Rightly or wrongly the huge enterprises will have to wrestle with the questions that Francois was asking. He was asking RAJ by the way and Raj is someone who by virtue of what his company is doing should be able to speak to these issues. So Raj please don’t be shy. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA MY intent is to publish a COOK Report that tells BOTH parts of this story. I do need to get paid to keep on doing what I am doing and I think it is incumbent on me to explore the current telephony morass of this so that my actual and potential customers can better understand the choices they face **NOW** while being inspired by your message that the next wave of change that sweeps all this aside is not too far off. Therefore, part one of the VoIP issue for me is what Raj and Francois and Frank and Henning and others are talking about while part two is very definitely your insight as to where with Skype and similar approaches this is inevitably heading. You are of course free to focus specifically on part two. :-) Menard: What I dislike about David’s view of the world is that it is incompatible with regulatory objectives of justifying forbearance upon a competitive supply of alternative access facilities. I do not have any business case for deploying wireless if the ILEC decides, can, is allowed or is not prevented, from giving away its lousy DSL services. I contend that VoIP will lead to more distributed, faster, more reliable peering between more carriers in a way that is not as centralized as with the current NAPs. For instance, I do not have the facilities to peer in Toronto if all I want to do is do VoIP over local ILEC DSL loops. Reed: Francois - in Canada the regulators may feel that they know what the customers need and what technologies should be granted forbearance. Menard: This is certainly better than the FCC for instance, who does not appear to have a clue about how to achieve sustainability in telecommunications competition. Reed: It’s not their job. Why should it be? Don’t obsess about the FCC. They are not God, cannot be, unless we make them so. They are pretty pitiful, just like the Wizard of Oz. In a democracy we construct our own bad guys and give them power by our fear and attention. I suggest you spend 4 days per week NOT thinking about the FCC, instead thinking about what you can build and building it. On the 5th day, you then have the credibility to tell them what really works for real users. Competition and Wireless Paths WiMAX Troubles Only in Vonnegut’s world of “Harrison Bergeron” does one measure or achieve sustainability by focusing on big winners and big losers, trying to manage everything into sameness. Any sufficiently large and *productively evolving* system has winners and losers at all scales (fractally scaled - lots of little losers, fewer medium sized losers, a few big losers). Competition at its most productive is a rough game, just like productive evolution. You can measure competition by the rate of failure (which tends to equal the rate of entry). When it’s competition for the attention of customers, it’s the most efficient filter we have of good ideas, because it seeks the optimum satisfaction of the humans who are the ultimate customers (the telecom industry seems to think that ILECs are customers for technology, and that devices are customers for bits). Ultimately what the entire information economy feeds on is the attention of humans to each other and their environment. The rest is hot air, self-justifying feedback loops that are accidents of the particular architecture we’ve evolved. When it’s competition to win on the “regulatory playing field” (which is where your arguments are always based, Francois), it’s one of the least efficient filters of valuable ideas, because it seeks the optimum satisfaction of the regulators (who aren’t the FCC, but mostly a small set of politicians, who largely groove on power over others). WiMAX’s fate (yet undecided, though I think it will fall into a miniscule shadow of what it could have been, just like Bluetooth has) proves nothing - that’s the mistake of synecdoche, i.e. the case 52 where a metaphor conflates an instance with one of its categories. Shockey: I’m increasingly interested in this. I have heard anecdotal evidence that WiMAX has some big problems with fog and other extreme forms of weather that is peculiar to the RF absorption characteristics of the spectrum WiMAX typically uses. Has anyone heard anything more about this? Reed: WiMAX has bigger troubles. It’s no different than cellular from the 50,000 foot perspective. It’s not opening a new market, but merely using the same technology to try to start a service that competes with a universally present service. The only advantage it might have is that it is not being positioned as a mobile service with roaming, so it can be rolled out incrementally. But cellular data services are easy to add. Somebody should study Christensen’s book again, and notice that wireless broadband need not enter the market in the markets that have already been cherry picked. 802.11 is a much better starting point - every computer user already HAS one endpoint of that system, and will buy whatever feeds their itch. 802.11 reaches the telephone poles in your neighborhood, if you put a bridge in your window. And if there are no telephone poles, it reaches your neighbor’s apartment or your neighbor who has a fiber with lots of spare capacity for now. That’s HFW (hybrid fiber(cable)wireless). A lot cheaper and more upgradeable than APON, EPON and other PONzi schemes. Cellular is already taking share from some wireline services at a rate that erodes the profit pool. There are lots of other wireless paths beginning to unfold. Cable competition (facilities based competition in HFC) is finally born in some regions, despite the fundamental socialist intuition that no economy can sustain 2 fibers to any home (a lot of hooey based on the old Big Lie that demand is inelastic for communications). Menard: I want to comment on David’s last three paragraphs above. The COOK Report on Internet May June 2005 First, I fully subscribe to your assessment that consensus (i.e. balancing the public interest against that of the interest of competitors and of the incumbents) is a pretty weak filter for new ideas. Still, the IETF revolves around the notion of rough consensus and running code. But what I want to point out is that there are cool new technologies like: a) ENUM which is set to put a new interface to the old SS7 who will not be implemented rapidly unless it is forced down the throat of ILECs by the regulators. b) wideband audio, which will be transcoded back to G.711 every time someone will want to send wideband audio to the customer of an incumbent, this again unless it is forced down the throat of the incumbents by the regulators. Shockey: Or the ILEC’s take it over as they have done in Poland and effectively kill it. Control of a communications service is directly related to the control of the underlying naming and addressing mechanism. In the Internet people can acquire their naming and addressing independently of the underlying transport or service provider which is not the case for either Skype or traditional PSTN service. Which is another way of saying that ultimately SIP wins in the long run. Reed: You are right if you mean that the service forcibly links the two. But remember that SIP isn’t a service, and its users (like Vonage) are trying to forcelink the two as we speak. SIP doesn’t win if its vendors wait for .e148 because they think the business is about PSTN. It’s in a new start-up’s interest (Skype) to decouple naming so they can connect to everything else over time. Shockey: Looking at Skype, AOL IM, MS-IM you will see that however viral the system seems, it ultimately runs into the wall that the control of the naming and addressing scheme is directly controlled by the underlying service provid- er. Separate those functions and you have the preconditions for a global versus a viral service. Governments remember have always controlled phone numbers though they use licensed carriers as the distribution mechanism. Why? Because control of the service is directly related to control of its naming and addressing. Menard: Second, regarding your agreement that WiMAX is a figure of speech, I do think that until such time as it is deployed, that’s pretty much all it is. I am deeply concerned with the (ahem, forget regulatory now) market economics which will lead to WiMAX being deployed in a manner that is sustainable. Third, regarding cellular, fact is that ILECs are the biggest wireless phone operations and that divestiture of their wireline operations is certainly not far away. But then again, wireless, at least in Canada, is a game of 3 players - I find no comfort in a vibrantly competitive oligopoly of 3 players, two of which are ILECs (Bell, TELUS) and the third of which is an incumbent cable carrier (Rogers). Earlier Reed: I grew up in a world where a business case involves the risk that someone will compete with you, and even price aggressively to get share. (I was responsible for several billions of dollars of product in that environment, and there were NO guarantees that somebody might not price in a predatory manner (well, Robinson-Patman, but that’s pretty weak protection). The ILECs are completely uncompetitive, saddled with ancient technology and inefficient business practices, without their regulatory protectors to protect them from competition. Menard: If they are incapable of competing, I wonder why WISPs are not expanding in areas where ILECs offer DSL services? Gill: That is because the services provided by wireless offer no compelling advantages in either cost or bandwidth/ latency that can overcome the inertia caused by the transactional cost to switch 53 to them. There has been no business case so far that I’ve seen. Caveat - I mean wireless for data that can compete with broadband cable/DSL plant. Reed: The ILECs may or may not still have some advantages, but don’t give me any foolishness about a business case having to have NO RISK from competitive response. Menard: The risk has to be that ILECs will not have price flexibility which allows to price their services on a promotional basis at a level that is below their incremental costs, which is the price floor used by competitors that can be used to begin achieving sustainability. Reed: Only in the crony socialist world of telecom would anyone believe that margins are a multiplier on costs. Margins are the difference between value and cost of the cleverest provider of that value. I happen to think that wireless solutions can cost pretty damn little, if you use smart technological designs. If the frame of your dreams is that the only answer is fiber through my rosebushes, well, you are going to have to beg for protection from the government so you can hire the children of the politicians into cushy jobs. Menard: Another day, another round being won.Ssee for instance this morning’s decision showing my regulatory prose in action: http://www.crtc.gc.ca/archive/ENG/Orders/2005/o2005-21.htm Reed: They call equity investment RISK CAPITAL. I prefer a world that rewards people who figure out how to get funded and deliver stuff end users want. Not a world that rewards incumbent managers and golf-playing heirs for sitting around and ordering their staffs to service their personal ego needs for status and importance. Impact of Smart Design on Wireless Cost It’s not the regulators who will enable connectivity. People want it, and will buy it at a price that some technologies COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA can meet with handsome profits. Wireless solutions can cost pretty damn little, if you use smart technological designs. Menard: Such as? How do you beat 3 mbps DSL for C$19.50 a month with any wireless technology? Gill: I am very curious to hear this myself. The numbers are confidential but I think given the right alignment, we can come close to that number in USD given current technology in the 900 mHz band using for example, Waverider gear. Can we beat three mbps for the same price? Not to my knowledge. David may know some vendors that can, I do not. David, anything you can disclose here? Reed: If the ILECs try to price DSL low, the fact that they are being killed by cellular and cable competitors will synergize to hasten their demise. Menard: Question? Fact? Fiction? I’m seeing no evidence backing up your statement. Gill: I’m interested in the reasoning behind this as well. Ballard: And who is picking up the cost of the Waverider CPE and the truck roll associated with this? Gill: It is in the model, with some assumptions such as return CPE if it doesn’t work for you, plus RF planning and prequalification, you can reduce truck roll. Far more costly on the wireless end though is interference. What most people don’t realize is that if a customer support call comes in from a user, you lose all margin on that customer for a good while and with high churn rates, it well may be for the life of the customer. The costing model is a fully loaded cost, including equipment writedowns, MRC for pole rental, power, and SG&A. Ballard: Interested to hear how this very low cost model provides payback. Gill: We are not sure, hence the experiment. Models are only so good, at some point the rubber meets the road and I’m a very big fan of empirical verification of Excel. Menard: You cannot afford a truck roll the stuff needs to go through leaves and it must work from your basement too, like Tim the Toolman Taylor says: WE NEED MORE POWER. That’s why I’m waiting to see what power Industry Canada will allow in Cognitive radio in the TV band, i. e. 802.22. How to Think about the Strategy of Power in Terms of Part 15 Reed: Excuse me, Francois. That’s entirely the wrong understanding, though if you are focusing on one link at the expense of all in a reductionist way, you can be forgiven. More power is almost always the last thing we need - you get faster computation by dropping the energy in each transistor, even though you get faster transistors if you raise the energy in each; the same thing holds true for radio communications. Radiated energy merely pollutes the electromagnetic field with energy at all the places that don’t want your signal. Not only does it waste the energy input, killing your batteries and contributing to global warming, it also does not improve the system performance as a whole for reasons that perhaps I can hint at below. Menard: OK, but do you consider that its fair for the FCC to limit ISM to current power levels and then compare the power levels that the cellular carriers are allowed to radiate against the power level that WISPs are being limited to? Reed: It might even be smart! The key thing about Part 15 (*not* ISM, which is the primary licensee of the 2.4 GHz band in the US, or Amateur Radio, which is the secondary licensee of the same band - “unlicensed” is at level 3 in the regulatory framework!) is that Part 15 must accept ANY other signals, and must also not interfere with primary or secondary licensees (aforementioned).There is no “right” that Part 15 devices have to transmit at *any* power level above zero. The reason Part 15 has worked is that it has encouraged innovation that provides 54 useful service despite such restrictions! Including the right of Amateur Radio Operators to shut down T-Mobile Hotspots anytime they can *prove* interference. But Part 15 has shifted the presumption of interference and burden of proof. In those bands, ISM, NII, etc., unlike the AM/FM/TV/Cellular bands, the real fairness arises because of a simple legal shift, not power levels. In the TV band, the regulations presume interference even when there is no television present that can hear the signal, much less have its picture degraded. Such “interference” is a fiction. Just as to prove burglary, you must prove more than that someone walked into the house. Earlier Reed: Ultimately, we need more network-scale coordination, cooperation, and sharing, less balkanization and focus on one greedy link at a time. The cellular radio industry has demonstrated continuous increases in capacity while dropping radiated power continuously over time by architectures that follow exactly this strategy. [Your brain and corneas thank you because although the old cell phones were barely safe, if we had increased power to increase coverage and data rate, rather than using better techniques, you would have to hold your cellular radios at arms length to be safe from cooking key proteins.] Remember that capacity of a link increases only as the logarithm of power, which means that to double data rate, you have to SQUARE the power! To double distance at the same data rate, you have to square the power, or if the curvature of the earth is taken into account, you have to raise the energy to the fourth power. And this ignores the other effects of putting out more power: First of all, putting out more power to cover more is a “one way broadcast” concept. Networks are interconnects, not one-to-many. It does no good to have a tower with more power, if the other end transmitting back does not. Second, the effectiveness of a network is not measured in ergs or joules. It is measured in the same units as entropy, not energy - bits (information is precisely negative entropy). The key to scaling information capacity is to manage the system’s distribution of entropy, which The COOK Report on Internet May June 2005 is largely independent of energy. Shannon showed that it’s based in the ratio between energy that has the relevant information and energy that does not. Once you have sufficient wireless activity, true natural noise is not the issue - multiplying every transmitter’s energy by an arbitrary constant does not change the information capacity one single iota, because entropy is independent of the energy. Or in other words, doubling every radio’s power output also doubles every receiver’s interference input, which means that the ratio of signal to interference stays constant. Similarly, halving every transmitter’s energy halves the interference at every receiver. A system of N independent units distributes information (i.e. negative entropy) as a result of decisions it makes in its components that correlate and respond to incident signals, routing the information from source to destination. The function of communications is only a redistribution of entropy, not a gain or loss in total energy. And computation only need cost energy when it increases entropy (read Feynman and others on “reversible computation”, which is an area that we are beginning to really exploit as we move closer to understanding how to design processors with more and more reversible logic) - communication decisions within the network do not necessarily increase entropy throughout the network - the key is to design networks that get closer and closer to changing entropy only at the sources and sinks. In optical fibers, we are getting closer to such limits - we send signals down incredibly long stretches of fiber without amplification by designing with entropy and not energy in mind. The key thing about a fiber is that it doesn’t change entropy except at the endpoints (solitons are nice entropy stabilization techniques). We have only started to have the scale to do that in RF networks. We are already able to translate microwave signals into optical domain without losing energy remoting microwave signals over optical fiber (read about microwavephotonics.com’s technology). was raised by Francois: Regarding working in your basement, a trivial example applies: just put a very simple RF repeater or translator in your house, and you have service in your basement. This is no different than the leaky-coax systems that distribute RF throughout tunnels and buildings that cannot be penetrated from outside. The net improvement in efficiency, measured in units of system entropy is huge. Why pump up the volume so a few deaf people can hear, when you can give them hearing aids or cochlear implants instead! It’s natural, but UTTERLY wrong, to think of power as “punching” through stuff. Adding energy to waves in electromagnetic fields just dribbles more energy and increase entropy (decreasing information capacity) in almost every case. This is true of RF, optical energy, X rays, and gamma rays. Power is Marconi-era thinking. Networking (especially *inter*-networking, where connectivity increases for everyone with every radio you add, according to Metcalfe’s Law) wins over raw power. Last week I Skyped a fella in Melbourne, AU who was using a WISP supplying pt2-pt 512 symmetric, and paying $49.95 Australian capped at 1GByte per month. Which, he cautioned, would not buy as many loaves of bread as $29USD. That said, his alternatives were dial up and sub-384k DSL undependable (his words), at a cost of ~$30/mo., with no two-way over cable coax seen on the horizon, yet. http://www.goldenit.net.au and http:// makeashorterlink.com/?L14A32D3A This morning I posted the Intel story (upstream) that included mention of a wireless offering in Minnesota of 1Mb/s unlimited coming off lamp posts and adjacent rooftops for 15.99 per month. And lest I forget, the muni in this case is supporting the service. If you want to tell regulators what to do, tell them to declare radios that don’t participate in networking enemies of the people, and make producing them an offense that carries a life sentence for the engineers who produce them. From an architectural perspective, which includes being able to accurately assess the feasibility of such platforms in a wide variety of circumstances, one needs to be able to predict what the shelf life (sustainability) of such services are before they are deemed obsolete, and what point solutions they aim to satisfy. At 1Mb/s, we’re getting close to, but not quite there yet, where one would want to be able to satisfy most residential so called triple play needs. Menard: If one day I become a politician, I will sponsor an amendment to the criminal code that makes plotting against innovation a criminal offense. Drop the CableModem and telco voice and keep the CableTV services, while picking up voice and Internet access over a Wi-Fi platform? An Almost Hopeless Complexity of Variables Involved in Service Pricing Wait for a marginally higher speed wireless service to come along that will include IPTV? Coluccio: If we’re addressing these issues on a global scale, much will prove to be local and terrain and nation specific. Allow me to post a couple of examples that bring out the economic realities being spoken of here, and at the same time address the disparity of pricing issue between wireless and DSL that 55 How long will analog FDM NTSC/HD services continue to dominate cable operator architectures and set top boxes? Will voice continue to present the same level of revenue down the road, as it appears to the operators to present now? Just a few points while musing to tail this topic off, but issues that I’ve been wondering about recently, just the same, COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA so I thought I’d pass them along here in response to Jere’s concerns over in-elasticity and see what sticks. Coluccio: I guess there’s more than one measure of separation occurring here, With regard to the different universes where VoIP falls into play. Stastny: I agree. This was a simplified approach. See below. Coluccio: I commented to Gordon several days ago, partially in jest, that large enterprises could do a great deal to mitigate their net-borne security concerns by allowing employees, sometime in the not too distant future, to use their own universal platforms, as you suggest, both for their voice and data needs. This way you take the risk off the employee’s desktop and stick it in his pocket, where it can do far less harm to the network. Stastny: Yes and no, General purpose devices are basically also endangered like Laptops, and everybody travelling with laptops knows the paranoia of LAN administrators against “privately” managed laptops ;-). So you will get in future your business mobile phone handed out crippled from the admins. Coluccio: Given the type of universal PDA Richard Stastny speaks of (which corresponds to what I was speaking of) this is not a far-fetched idea. But the point remains that the center of gravity of VoIP decent- to higher- margin revenues for service providers will not be at the end points of consumer lines, where many of those will be delivering services that are virtually free at one point, but the more lucrative services will be embedded in enterprise and in governmental phone works for some time to come. Stastny; I also agree: There will be basically three types of “Service Providers” 1. Enterprises 2. Service Providers or better hosting providers (as now for email and web-hosting, etc) 3. DYI - you may buy and run your VoIP Server at home, like an e-mail server or web-server. This is now only for freaks, but more and more people run networks at home (more than one PC, media server, NAS for videos and photos (also for backup), control for networked appliances, so why not run also your own communication server. Normal people will buy them at the supermarket (as now PCs). Freaks will also buy them there, but reprogram them, because they all will run on Linux or MS. I read today in a computer magazine a story how to upgrade an out-of-the-box NAS (network attached storage device) by moving the Linux OS to the disk to act as a media server. And like with e-mail, many people will use both the office communication server and private communications server, either DYI or outsourced Reed: Even in the latter enterprise scenarios, however, I would not minimize how quickly employees on their own will move over to using their personal devices 56 instead of slogging over to their desk to pick up a tethered, black telephone set. All the evidence I need of this is to observe how, in my own home, both I and other members of my family automatically default to using our own cell phones, when there is a land line sitting in its cradle not more than a room or two away, or when the land line-wireless unit itself is sitting on the coffee table side by side with the cell phone. Stastny: I fully agree, it is the same with me: In the office I use my (office) mobile phone just for convenience, because I have the numbers stored there (only for longer calls e. g. conference calls I use the office phone - mainly because of the speaker and the mute button). At home (I have four kids and all have a mobile) I never pick up the fixed phone, because it is not for me anyway. The kids still use it for outgoing calls because I have to pay. ;-) My next gadget will be a dual-mode mobile phone (e. g. the already mentioned Qtek 9090 to get also my Skype and SIP call delivered on the same device, at home and everywhere in the world. So what will providers deliver: access and hosted services - I still want to have a voice box not dependant of my home network if crashed by my kids. The rest will be products you buy in a shop (on the Internet). Skype Seen as "Instant Voice-Integration" of Multiple Forms of Communication into Broadband Based Collaboration Highlights Improved Audio Codecs, P2P Architecture, & Other Features May Push Skype like "Instant Voice" Softphones into New Areas James Enck Explores Possible Impact on Wireline and Mobile Carriers Editor’s Introduction: James Enck is the European Telecom Analyst and Global Telecom Strategist at Daiwa Securities SMBC Europe Ltd. James is based in London. On the welcome page of his well-known EuroTelcoblog he states: “In July 2003, I began publishing an email blast called EuroTelcoblog as part of my work as an analyst at Daiwa Securities SMBC Europe.” He also explains that his blog allowed him to give some real expression to his desire to “observe how disruptive technologies interact with and ultimately subvert rigid legacy structures.” I interviewed James on Friday March 4, 2005. COOK Report: What are the most interesting aspects of Skype, or other P2P voice communication programs that you see at the moment? Enck: Some of the implications of presence for Skype are going to turn out to be very interesting. There are stories of using it as a kind of open microphone between two physically separated but otherwise emotionally very involved people. You may have also seen Martin Geddes' blog description of how he and his Lithuanian wife may leave an open Skype connection to the grandparents in Lithuania on all day long. It permits verbal communication between grand parents and the grand children on an "as convenient" basis. It’s a bit like having an intercom open to the next room. Future Possibilities: Improvements in Presence and in Audio The audio aspect is open to improvements in interesting ways. I have seen reports from Stuart Henshall that he has heard that Skype is working on spatial positioning of voices in a conference call. There is always scope for something like haptics where you can project some element of a person's physical presence such as a heartbeat to come into play here. In the grandparents’ scenario, the sense of being there and being able to listen to the kids play as opposed to having five minutes on the phone with them is worth thinking about. This of course becomes possible if the technology works and as long as the service is not metered. Alan Duric is the CTO of Telio, not the one that every one is talking about in the States but a company that is basically a Norwegian Vonage. Duric used to work at Global IP Sound where he was one of the authors of the Sype audio codec. His description of some of the things that are in the development pipeline with broadband audio codecs is that he says for the future it will be like the difference between black and white TV and color. This is where he sees services differentiating themselves once price is no longer an issue. The thinking seems to be that market share will be determined by whatever platform delivers the best sensory experience. This is an area where Skype already is quite far ahead. COOK Report: Certainly visual presence can’t be far behind? All manner of web cam devices are becoming very cheap. Therefore, it probably will not be long at all until Skype or its competitors do video as well as audio? As monitors get bigger there will certainly be room for a video window. As you move all kinds of presence to the edge on an individual basis, this should drive increased demand for bandwidth and for filling up fiber. 57 Enck: Yes - all of what you suggest is absolutely true. I think what Duric is saying is indicative of a growing realization that use of peer-to-peer audio communication softphones or software like Sype is just fundamentally different than products based on SIP that must use ordinary hardware telephones and be able to connect effortlessly to the PSTN. You have on the one hand what he is saying and on the other hand rather negative responses coming from the SIP folk and the SIP --using hardware community. This is indicative of the Skype’s having driven a wedge down not only the proprietary versus open standards issue but about what the very nature of the service is. Zennstrom has never made any bones about the fact that Skype was originally basing its interface and user experience on instant messaging adding in effect instant voice to instant messaging. They were never trying to recreate the telephone experience. COOK Report: This gets back to what David Reed was saying last December on the mail list about Skype not being an IP replacement for the telephone but being designed for strongest appeal to those living a communications centered life. One has to really re-orient one’s mindset in approaching this. This leads me to ask about their use of the Global IP Sound codec. How closely are they tied to that? They have to be thinking about how they would add features? Enck: My understanding is that Skype has simply licensed the codec. I am not sure how the payments are structured namely whether it is based on number of users. My understanding is that they have COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA to make regular payments. Either quarterly or semi-annual. I don’t think they are in anyway locked in but they have so many clients installed that, if they wanted to do something dramatically different, then everyone would have to download a new Skype with a new codec. It might take a bit of a while to roll over their entire user base, but it would not I think be an impossible task. COOK Report: Stuart Henshall mentioned a new 3D codec coming from Japan. But other than Alan Duric in Norway what do you hear about new codecs? Would 3D audio imply multiple speakers or probably just a good head phone? Enck: With headphones, the 3D would imply that you could place people in a 360 degree stereo spectrum as to the positioning of their voice. With a conference call you would get the image of people actually sitting around a table. I would think in a conference setting this would be very useful in a situation where essentially what you have now is 10 mono channels over lapping. It would make it much easier for people to identify who is speaking and in that sense I thing they could learn a lot from some of the gaming enthusiasts. In this area there are a couple of products - one is called Teamspeak and the other is Teamtalk. These products offer a way for multi-player on-line gaming administrators - basically the people running the server - to impose some moderation on the process of running the game. I believe you can see who is talking as a part of the gaming character on your screen. It is a very advanced kind of presence that they have built into this thing. Within Skype, for conference purposes, perhaps you could have something like this built into a future version so that you could see a representation of a meeting with a representation of the person speaking his name and an indication of whether that person’s audio channel was active or not. In an even more simple form, if Skype lit the icon of the speaker and you could couple it with some sort of stereo placement of voices, it might be really interesting. There are lots of variations there that are possible. I sent a note the other day to someone I know at Skype asking why didn’t they put in a record function so that people could easily record their Skype conversations or conference calls? If you think about it being used in an enterprise where there are compliance issues or where you want to archive your discussions, having a record function would be useful and it couldn’t be that difficult to do. within Accenture. The reading I was given is that they are currently interested enough in it and want to see where it will take them such that there is no effort made to stop it or restrict it in anyway. The folks I have talked with stress that, while they are not going in clandestinely and using this behind a client’s firewall, it is something they feel comfortable enough to use within their own organization and that they are not discouraged in anyway from doing this. The response so far from Skype has been that we haven’t seen that much value to us so far in doing something like that instead of other things that are higher on our list because they appear more immediately attractive. COOK Report: What do you hear about the dangers and risks of a peer-to-peer system like Skype? Enterprise Concerns about Skype Encourage Competitive Products COOK Report: Perhaps this is because of the very negative reaction that Skype gets from enterprise security people? Would you describe what you hear from enterprises as to their use or non use of Skype? Enck: I wrote a piece not long ago on Accenture. It turns out that this piece flushed out two other guys. They have project teams working in Pakistan, and other teams in Amsterdam, Madrid, Vienna and London. First of all within Accenture itself I have confirmation that top managers and even board members are aware that this is going on. While there is no official policy pro or con, they have never said to their employees that you cannot use Skype. Apparently at least one senior board member of Accenture is a Skype user himself and does so within the business. Now when they are out at client sites what they do apparently is set up their own free-standing Internet access point. They would work through the local telco to construct a VPN that would enable them to set up their own access node independent of the client’s network. Or perhaps they would just bring in an independent DSL line. Therefore, the client network is never at risk. It is Accenture 58 Enck: While I am not a security expert, I have certainly noted Melissa’s concerns about an open Port 80. I have also noted the opinion of Dimitry Goroshevsky at Popular Telephony http://www.peerio. com/ and the folks at Nimcat Networks in Canada http://www.nimcatnetworks. com/ that any IT manager who turns loose an application that can punch a hole in a corporate firewall needs his head examined and that is why they have built into their application an administrator role where central control is required to switch functions on and off. COOK Report: So isn’t this another way of saying that what is going on here will probably evolve into new software packages, new devices and new companies? Right? Enck: Absolutely and I think that there may be future iterations of Skype that embrace enterprise security concerns. It can’t be something that is perceived ass a renegade product. They will realize that if their product is to find widespread enterprise use, it has to be controllable from the IT manager’s view point or it will be seen as an unacceptable threat no matter what the perception of benefits by the other folk in the enterprise. We are already seeing some activity from another Norwegian company called Paradial whose leaders are veterans of the old Ericson Voice over IP research labs. They have something called Real Tunnel. Basically it’s a firewall traversal product for MSN Messenger that I have written about a couple of times. It pro- The COOK Report on Internet May June 2005 vides a Skype like functionality for MSN Messenger. It works like a treat and has no trouble getting out. www.paradial. com/documents/RealTunnelTM.pdf It’s a SIP application where you are log into a proxy server and where the purpose is to deliver P2P like functionality to MSN users who find themselves behind firewalls. You get pretty high quality voice. You can do video. There are some other companies with similar application developments. One in the UK called Ridgeway Networks was acquired last year by Tandberg the conferencing people in Norway. Alcatel and Cisco have made similar acquisitions in the past year that seem to have been driven by the perceived need - namely to bridge firewalls in such a way that desktop conferencing can be safely allowed. http://www.ridgewaysystems.com/products_ipf_faqs.aspx I think the notion of enterprise use is being validated but that the central authorities also want more control than Skype currently offers. COOK Report: And with Skype’s success there will be other people stepping up to the plate? Enck: Yes. Goroshevsky and Popular Telephony COOK Report: Let’s come back to Goroshevsky. I have seen from reaction on my Network Economics mail list that he has little credibility because he has been announcing products for many months and has not delivered. What is going on here? Enck: Over the past two or three months their press releases have made it clear that they have a working relationship with Texas Instruments (TI). COOK Report: And TI would have denied Goroshevsky’s assertion were it false? Enck: I think that indeed you can count on that. It looks now that TI has ported his software onto its DSP. Given that TI is there I would be surprised if Broadcom is not also involved. My surmise is based on the fact that Broadcom is a financial investor in Nimcat Networks which is Goroshevsky’s principal competitor. Also I know someone in the product and development unit of a large European electronics firm. These folk have actually done trials of the Peerio application in their laboratory and had good things to say about it. COOK Report: The software resides as firmware in a TI Digital Signal Processor (DSP) chip then? Where would one find this chipset? Enck: It would be in the desktop VoIP or hybrid kind of phones that are coming out. The ones that they, in late February, announced from Grandi (a Chinese firm) is a good example. That phone has a PSTN connection, two Ethernet jacks and also a SIM Card slot. Popular Telephony has signed agreements with about 6 different makers at this point. They have Gateway people involved. COOK Report: Their software loads into the DSP as firmware? Enck: Yes. It is firmware. Middleware or whatever you want to call it. Theoretically at least it can sit on mobile handsets or on virtually any other device you want to think about. COOK Report: How would it work? Enck: I think it would be conceptually similar to some of the clients developed around Skype by third parties. For example there is a product called the DualPhone, a DECT phone for Skype <http:// fa86dd8e8eff5070c1256f1c0040dee5. dualphone.net/>. It is a cordless phone that is available in Europe. It is both a wireless PSTN phone and one that affords you a connection to your PC and to your Skype contacts. You can access both and choose what kind of call you want to make from this single handset. Conceptually that is how it would function. What the mechanics at this point are going to be I am not sure. Whether there is to be a purple button with a “p” 59 on it to launch Peerio for example? Or whether it is in the numbering or in how your address book is structured? COOK Report: But presumably it could also be distributed as a software client? I understand that like Skype it is peerto-peer in its structure but presumably it is how the “peer-to-peerness” is implemented that is different? Enck: Yes. I think that is a good example of the difference. Peerio is a bit of a black box. I had a conversation with Bram Cohen, the guy who wrote Bit Torrent. Although he was not aware of any of the P2P voice systems I gave him a description of what Dmitry was doing and said that I had asked Dmitry whether Peerio was similar to any other P2P application? Dmitry had replied "Well, yes and no." Conceptually it was similar to something called Chord. Later I read some academic papers describing Chord but did not understand them very well. However when I mentioned Chord to Bram Cohen, he immediately said that he understood what Goroshevsky was doing. For this I conclude that within P2P coding circles, this is an architecture that is actually understood. The folk from Nimcat networks also seem to understand this. They, by the way, have given us an in house demonstration where they brought three handsets in. They set up a miniature LAN in our offices. The person responsible for the demo configured one phone and assigned the other two identities that were basically extension numbers. We switched the phones on and watched as they used the network to discover and configure each other. We were up and running within a couple of minutes on a demonstration where they went through all the standard PABX functionality. We forced one phone to fail. The purpose of this was to ascertain the fate of all the address book information and other user data that would normally reside in the phone or on the central server, but in this case is actually cached within the network. It worked. We plugged the phone back in and it recovered all its data from the network. There are reasons to believe that this fully distributed architecture COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA does work. Unfortunately I would have to say that all the proof I have to substantiate this comes from Nimcat Networks rather than from Popular Telephony. At least the things that Dmitry is saying seem to be substantiated by his competition. COOK Report: But you have talked to some folk who tell you they have seen working trials of Popular Telephony systems? Enck: Yes. And if I look at their agreement for the French phone, I believe those are due to ship before June. If they are really playing with people it will become apparent pretty quickly. I can understand people’s impatience, but I have a hard time believing there is nothing there because for there to be nothing there would be the ultimate suicide. What would Dmitry have to gain from releasing a bunch of press releases and then producing nothing? COOK Report: And someone from Texas Instruments would, presumably, have sat on him long ago for claiming TI’s involvement? Enck: Definitely. My view has been that TI is hedging its bets on an interesting looking technology. But TI is not the only one involved. Some European consumer electronics firms involved in the hybrid set top box arena are involved at least in looking at the technology, although I am not aware of any decisions at this point to actually put them into equipment. COOK Report: A lot of folk had better pay attention because we see now that broadband data networks are morphing into telephone networks. Enck: Indeed. COOK Report: A question about Mike Volpi at Cisco who as of last September 15 joined Skype’s Advisory Board. Volpi has to have a pipeline now into Skype’s development plans and one had better believe he will be watching out for things that impact Cisco’s interests. Are you aware of any other firms whose executives have advisory type positions with Skype? Enck: You are asking good questions and I am afraid I don’t know the answers. I know Zennstrom, having eaten lunch with him a few times and I know his principal backer - Draper Co. in California. Now that you are asking, I am surprised that this information is not really public. Skypeʼs Cellphone and Wi-fi Direction COOK Report: How would you describe the wireless wi-fi direction in which Skype appears to be headed? I think Stuart in his conversation with me was suggesting that Skype viewed its value more in mobile technology than in convergence with the PSTN. Enck: I think that is correct. Certainly in Europe the mobile area is where a lot of the arbitrage opportunity for subscribers exists. The reason for this is that if you are traveling around Europe, you pay very heavy charges for roaming. This is an area that costs businesses a lot of money and where they would be very keen to make savings. Based on anecdotal evidence I get people from the States seem to be happy to go into a T-Mobile or BT commercial hotspot and pay seven pounds for an hour because in that hour they can make forty dollars worth of international calls over Skype. Very worthy of attention in this area is the announcement yesterday (March 3) regarding Broadreach. They are huge in the UK. They cover all of the London train terminals, as well as all the MoTo roadside truck and car stops that are really somewhat like miniature shopping centers. They claim that 500 million people is the annual total of the daily visitors to all their locations, although obviously people aren’t going to be carrying their laptops on each and every visit. Broadreach is basically a wi-fi managed services company that is hired by the railroads to wire up the train stations for wi-fi and so on. They actually do give services to a chain of coffee shops called “EAT”. They manage and connect to 60 the Internet the networks of the various enterprises that buy their white label services. They seem to specialize in white label wi-fi networks for retail and travel organizations. One of the companies I talked to recently that does work in this area was getting about 350 to 400 pounds sterling per month per site. They have a little Linksys wi-fi node and DSL connection and come by and check it out every few days. COOK Report: But a London train station takes much more substantial equipment? Enck: Oh yes. You can bet on that. Meanwhile some of the locations are quite desirable and this will expand enormously the Skype friendly locations for people to work from. The release was worded very strangely and I am not precisely sure how the use arrangement will work. COOK Report: I understood it to be that you would be able to connect into the net in such a way that your Skype client could function normally at no cost but you would be unable to send and receive email unless you paid the hot spot. Enck: I agree that that it how it sounds. I am not directly sure of the mechanics, but I assume the network is set up from a technical point of view to recognize Skype. I don’t know if it would be done through port scanning or if you would have to make yourself known to management as a Skype user. The Skype question aside, in many UK hotspots, usage is geared to purchase. You make a minimum two pound purchase and get 30 minutes free use. How it’s going to work in the train stations is not entirely clear. Whether in the lobby or in pubs or in coffee shops? I don’t think it will be train-station-wide because there are other white lable wi-fi network management firms besides Broadreach that deal with train stations. BT has some points there, and also at airports. Editor: Thanks to Sebastian Hassinger for this additional note: http://www.benhammersley.com/weblog/2005/03/03/ snow_and_skype.html The COOK Report on Internet May June 2005 “Meanwhile, one terribly interesting thing to arrive in the Press Release Inbox this morning: Skype have signed a deal with Broadreach to allow Skype to be used free over the 300 or so ReadyToSurf wifi hotspots in the UK. Broadreach’s ReadytoSurf(TM) network of locations has a combined footfall of over 1bn per year and includes major brands like Virgin Megastores, Eurostar, Travelodge, Moto, Little Chef, Virgin Trains, EAT , Choice Hotels and Quality Inn and major railway stations including all the London terminals.” [end of press release – Hammersly concludes: “You’ll still have to pay to get web and email, but the firewall is unlocked for Skype traffic. You hear that? That’s the sound of the future.” COOK Report: Malcolm Matson claims that people in the London office of Skype have said that Skype will be loaded into wi-fi phones before Christmas. Will these phones be dual use in the sense that flip a switch and they operate as cell phones as well? Enck: I believe the Motorola agreement was exactly that. It looks to be on a class of phones that have a handoff capability between GSM and wi-fi. I don’t know whether it will be on a best effort basis or whether it will be the user who flips a switch. I think the carriers will obviously have a problem with this should you be able to do a wi-fi call rather than a Vodafone one. But I think many enterprise customers would VERY much like to have this control. COOK Report: How far are we from these dual use handsets being on the market? Two or three months? Enck: I think that is a about right for the enterprise market. And perhaps another two to four months beyond that for the general consumer market. My reading of the handset makers is that they are scared. They have been the victim of the carriers for too long. They have seen the carriers hammering them down for too long. They are begining to see that the carriers have a limited shelf life so they have to begin to ask whether they want to continue to be a slave to this dying industry? Or do they want to get in bed with an application developer like Skype. You can see it now not only with Skype and Motorola, but you can also see it with Microsoft Nokia announcement that shocker everyone so much a few weeks back about allowing direct connectivity between Windows XP and Symbian based handset in order to share music. This will immediately eliminate a certain amount of carrier revenue. Who is going to pay a dollar fifty for a song download over the network when you can simply down load to your I-tunes and the upload it to your phone? Nokia signed an agreement with Macromedia developers to allow more independent applications coming in. They have also opened up a Python forum. Python is apparently well tailored to writing things like SIP applications. Bit torrent is written in Python and the fact that they are opening up the Nokia platform to this kind of programming language that is associated with wild disruptive open source technology is a further sign that the handset makers don’t want to be enslaved to the carriers. I think Motorola was really sticking up its middle finger at its traditional customer base and saying "look we want to carve out a sustainable stake in this sort of value chain because we think your traditional model may be in trouble and we don’t want to be on the wrong end of that." I think also that the Motorola guys will have seen the Taiwanese makers HTC, or High Tech Corporation trying to court Skype. They will have also seen the Skype I-Mate announcement and find themselves getting very paranoid. COOK Report: Are you seeing anything from Skype about how soon they are going to be out on Symbian? [Editor’s Note - Global IP sound has announced a version of their Voice Engine codec for Symbian during the week of March 7.] Enck: No and that is an important question. If they really want to cover the mobile front, they do need to get it out there. 61 Skypeʼs Business Model Have you ever looked at the Skype job listings page? That can give you some ideas about where they are headed. http:// skype.com/company/jobs/ And then you will see an interesting page for business development manager listings for their London office. http://skype.com/company/jobs/london/index.html#job-bdmm Business Development Manager - Mobile Devices The Skype Business Development Manager, Mobile Devices (BDM-Mob) shall play a key role in establishing, developing, and managing commercial relationships with key Skype device partners shaping the future of the mobile industry. Business Development Manager, Retail The Skype Business Development Manager, Retail (BDM-Ret) shall play a key role in establishing, developing, and managing commercial relationships with key Skype consumer electronics retail and distribution partners. Business Development Manager Computing The Skype Business Development Manager, Computing (BDM-Comp) shall play a key role in establishing, developing, and managing commercial relationships with key Skype device partners, shaping the future of the computing hardware and peripherals industry. Business Development Manager - Mobile Services & Technology The Skype Business Development Manager, Computing (BDM-MST) will play a key role in establishing, developing, and managing commercial relationships with key Skype mobile and technology partners. Business Development Manager - Internet COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA The Skype Business Development Manager, Internet (BDM-Int) will play a key role in establishing, developing, closing and managing commercial relationships with Skype online partners around the world. to the sale of KaZaa to Sharman, That story in my mind was not about free speech or rebellion but was about selling the entertainment industry something that could be used something that could be used to distribute content. Basically they are hiring in all the key market verticals that you would expect them to be approaching. This list used to be a lot longer. They have obviously filled a lot of positions. They used to have one that was for embedded products which is similar to what Peerio is trying to do. So clearly there is stuff happening there. Mobile is clearly one of their priorities and, according to these listings, perhaps an area where they haven’t filled in all the gaps. COOK Report: Unfortunately the entertainment industry didn’t get the intended message. COOK Report: What else can be said about the Skype business model at this point? When Motorola puts the Skype client in their handsets they presumably have to pay Skype? Enck: I would think so. I think a large part of what they are up to is really licensing. From what I have heard them saying, I think they have been shocked and surprised by the uptake of Skype Out. I don’t think that it was ever a part of their business model to try to make money from termination to the PSTN. I spoke to them for the first time in the fall of 2003 - a couple of months after their launch. It was clear at that time that it was never an integral part of their plan. It was n ice to have but they didn’t need to have it. They are having a lot of painful customer service issues with it and this may be one reason - namely that it exploded in uptake far beyond what they had envisioned. [Editor’s Note: On March 11 the Skype web site reported a million Skype Out customers for the first time.] Income will come from licensing, from Skype-out (and Skype-In beta tested on March 10th for the first time) and given where these guys came from, I certainly imagine that content will play a role. Given the KaZaa thing with what I would call phase one of the peer to peer story that runs from Sean Fanning and Napster Enck: They didn’t. There is a book about what happened to Sean Fanning when the labels sued him and shut him down. He just couldn’t believe it would happen. He seemed to have thought he could sell Napster to Universal music and retire as a teenage millionaire. Instead he spent a long time in court and Zennstrom is apparently worried about getting served so he doesn’t come to the US. A Content Distribution Strategy? My assumption was that there was always some sort of content strategy built into this once they can get big enough that they have several million users on line world-wide at one time. Especially if you look at what is in place to try to incentivise people to become more responsible file sharers. For example with someone like people with Weedshare where they set up a system designed to incent people to share content and if it results in a sale they get a cut. www.weedshare.com/ It becomes a limited pyramid marketing scheme. This was tried last summer by Heart if you remember them. They used it to sell their Comeback album and using the Morpheus file sharing system they apparently sold more copies than I-Tunes did. COOK Report: Now I am getting your point. Skype already has a file transfer capability and, with Skype In and Out, it is putting accounting capability into place. Consequently Skype has the basic infrastructure by which it could sell music or take a percentage of other file sharing revenue. And if you have some kind of content that you want to make available on Skype, you pay them a fee for letting you distribute it? 62 Enck: Potentially yes. Depending on how the mechanism works. Weedshare is built around Microsoft DRM and I am not sure that I believe in DRM. But if it does work what they would be selling is the social or viral nature of the Skype network. For example at the moment you are able to transfer files of up to two gigabytes - directly, person to person. I wrote about this and Zennstrom was unhappy. He thought I was insinuating that they were putting together a file sharing network. He rightly made the point that it was not file sharing because there was no directory - that is to say no way to query for desired content. There is certainly however the potential to directly transfer the content from one user to the other. Now when I threw this suggestion at one of Skype’s development people, he replied that there was no conscious intent to do yet another file sharing platform. Rather he said they wanted Skype to have the same capability as other instant messaging platforms in order to be competitive. But I think this is where they are coming from and that they may have a dormant element to their strategy that they are not yet fully aware of at this point. But you have to ask yourself if they really have the 28 million users they claim and 80 million downloads on March 4 at some point this is going to grow big enough that someone is going to say "Gee- how can we use this huge platform to sell our product? Or to get our message across or whatever?" At that point - Skype is the gatekeeper. I have no proof of these assumptions - but they seem logical. COOK Report: What we are talking about then is a really well entrenched peer-to-peer system. The ability to stop or root out such a system as long as access to the Internet is not licensed and controlled along the lines described by the author of the Digital Imprimatur is generally nonexistent? To try to control such networks would render so many other things inoperable as to effectively destroy the Internet? The COOK Report on Internet May June 2005 Enck: Yes. I don’t think anyone can root it out. It is here to stay. That is one of the things I like about it. There is no reasonable way that you could gain control of what it does. If you look at industry trying to create alternative products that will do what peer-to-peer networks do and create what people like about them, then they have a really lousy track record. In my view practically everything that has happened in Internet/telecom product development has been an accident. Email. SMS. These all found uses that were not intended by the developers and they have subsequently become massively transformative consumer technologies. I think instant messaging has been another case in point. What has Yahoo, MSN and AOL accomplished? What was that all about? What was the master plan behind it? It generates zero revenue for the telcos. It has caused all sorts of problems in enterprises. But they have also found ways to use it for their own ends by putting it safely inside of sub-netted playpens blocked off from undesirables. COOK Report: What are your views on Skype’s market place position? Do they have such a head start that no one else can overtake them? Skypeʼs Market Place Position and Prospects for More Controllable Competitors Enck - Being bigger than all the other VoIP implementations in the world combined it is looking pretty formidable. Yahoo Japan would be the closest in size. But what have they got? Five million? And Yahoo is an access-based service so you are stuck with an IP phone in your house. It simply doesn’t have the same feature set as Skype. COOK Report: What you mean is that it is rather like Vonage or Lingo? Enck: Yes. Exactly. Then you get down to the next level with firms like Vonage and Iliad in France. Time Warner Cable and so on. At this level we are still talking about only a few hundred thousand users. You put all of these together and you still get a number that is smaller than the registered user base of Skype. We are talking about over 2 million concurrent users at any give time these days and that is a helluva lot of people. COOK Report: What I am beginning to realize is that the feature set of things that can be done with a P2P instant voice program like Skype are by no means exhausted. But still Skype seems so well embedded that the competition’s feature set would have to be really stunning to get people to switch? Enck: Quite true. Likely the only way would be to take an entirely different approach. There is a company called Voipster in the Netherlands I have been tracking that had gone rather quiet. But I reestablished contact with them a couple of days ago. www.voipster.com/ They have a similar background to Skype in the sense that they are very small and have developer in Estonia. They were using the Global IP Sound codec but they dropped that in favor of something that uses less bandwidth because they are trying to get a start in some emerging markets. Their goal is to be an invisible partner to carriers or ISPs. Or potentially to people in the online world whether that is retailing or services. They should becoming out soon with exciting announcements. Their architecture is very similar to Skype but they allow a level of control at the administrator level with the set up of their systems that will be a relief to enterprise security people. COOK Report: In this sense they might be the software that gives Google working telephone icons for parts of its emerging local service activity? Services like this that could be used and administered by third parties rather than just out there like Skype might be more interesting to Google? And in the third world are if you can get by on less bandwidth, you have a lot of possibilities. Enck: Yes to both assumptions. I was talking today to a guy who works for a European carrier who is a VoIP specialist and he was talking about a new codec that has been developed by ECI 63 the Israeli concern which he claimed functioned on something like two kilobits per second. They claim to compress by a factor of 12. He says that the next time someone claims that their codec can function in 22 kbs, you just say that is no longer relevant. This is definitely something I have to check out because if it is the case then we have Skype over narrowband. Editor’s Note: On March 14 James added I think this URL may be what he was referring to http://www. veraznetworks.com/products/Digital_ Compression_Multi_product.asp COOK Report: So where is this going? It seems that bandwidth like electricity allows you to do a lot of things. And that bandwidth therefore is just a commodity. Enck: You’ve got it. COOK Report: But the LECs with their ability to insist that you buy bandwidth from them at their prices are an impediment to any progress. And they are stronger in the US than anywhere else it would seem. What does it look like here from your perspective in the UK? Enck: I agree you have a big problem and I followed last year the ups and downs of Utah’s UTOPIA project. They had to suffer and enormous amount of alarmist flak from Qwest and Comcast. I think the US is an exception in this area. If you look at Sweden, you will see how Stockholm pioneered the idea of a municipal fiber utility with Stokab. The Competitive Broadband Situation in Europe Now turning to the Netherlands, you really have the next generation of this at work with the Amsterdam municipal government trying to push through the initial phase of a project where when completed every home and business in Amsterdam will be connected by fiber on a carrier neutral platform that will be housed in a network operating company which is set up by the municipal government and is partially funded by the fees it generates from service providers. In COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA this sense it is very similar to the Stokab model. COOK Report: Kees Neggers with Surfnet6 and Gigaport in the Netherlands has certainly prepared the groundwork for this expansion. But the ideology here in the United States that says anything done by a government is, by definition incompetent, is doing us considerable harm. The direction of the technology is pushing broadband more and more in the direction of becoming a central economic utility - as vital to business activity as roads, water, sewers and electricity. Perhaps with more Skype like applications and the new things underway at Google, it will begin to dawn on people that broadband is more than just faster down load of email and web pages? I went to broadband only because I wanted long distance telephone arbitrage with Vonage but now that I am there it makes me see things in entirely new ways. Enck: Agreed. I think what you see now in Europe is that the carriers see broadband as a new revenue opportunity with the ability to charge for overage on monthly usage caps. What they are doing is putting a usage cap on their prices. The minimum BT product at the moment is capped at one gigabyte - but for heaven’s sake that is a less than a single feature length film. What is that in music terms? Four hundred songs? It’s a joke. What they want to do is sell you incremental bandwidth for a charge. Telenor which is one of the most forward looking companies in Europe tried this idea two years ago and quickly scrapped it because of overwhelmingly negative reaction. But all the telcos in Europe are looking at capped products. They sell you cheaply a basic connection with a cap or you can upgrade to the so-called unlimited plan. Unfortunately people then find that those have limits of 20 to 30 gigabytes. (http://www.bt.com/broadband/bb_info. jsp?targetSection=packages) An aside: Martin Geddes made an interesting post the other day about how people who are on bandwidth capped packages may be putting themselves at financial risk if their PCs happen to be supernodes on the Skype network. However in places where cable TV is making inroads, the cable companies are saying to consumers we will double you band width speed or double the cap for the same prices as the telco. This is hardly satisfactory. I see it as a feeble attempt to circle the wagons. You could really smell the fear in the Netherlands the other day when KPN announced job cuts of 8,000 positions over the next five years. This is 28% of their total work force. Given that about a quarter of the work force is in Germany where they have a large mobile network it means that about half of their domestic work force is on the chopping block. COOK Report: What will happen then? As carriers shrivel away municipal networks will fill the gap? Enck: In some areas I think this will be the case. But in southern Europe the carriers seem to be quite strong. Companies like Telefonica and Telecom Italia seem to be quite strong. There is much less competition there. In fact in Italy there is zero cable. France on the other hand has been a real wake up call because in less than two years it has gone from a point where the incumbent controlled the entire broadband market to the point where unbundled lines total 25% of the ADSL in the country. For 29 Euros a month you can get 15 megabit per second ADSL bandwidth plus TV and free phone calls within the country. If you want to compete in France you have to come in at under 29 euros per month which is really hard. I think that the environment will continue to vary from country to country. Germany is still a long way from being competitive while the UK is moving towards being hyper competitive. France is there already as is the Netherlands. COOK Report: What is happening outside of Amsterdam with other municipal networks? Enck: There are other networks being built and they are not all municipal. Some of the drivers of this process in the Netherlands are housing corporations. 64 The rates of home ownership are a lot less in continental Europe than in the UK and US. In Sweden and the Netherlands a lot of residential property is in the hands of private or quasi private and very highly regulated housing companies. What is happening is that they have an aging population and a competitive housing market and they want to be able to offer an extra amenity to their customers that is not only broadband but can claim telemedicine and security monitoring. There is a property company in the UK called Land Securities that is positioning itself to string fiber to all new housing developments in the greater London area where the government has committed itself to encourage hosing development. They will control the last mile and offer a fiber connection almost as the feature of a home. I think we are going to see more developments along these lines as one more source of pressure against the incumbents. But in places like Ireland it is going to be a long tome before you see that level of competition unless it happens in wireless which is the other wildcard - whether it is mesh or Wi-MAX or something proprietary like Flarion. COOK Report: How would you sum things up at this point? Enck: It is a ball of confusion! I think Skype has been an inspiration to watch. I was lucky enough to stumble on it a week or so after they launched and started to write about it. I think some people here thought I was crazy but some other clients clearly grasped the meaning of this from Day One. We are talking here about institutional investors who have people’s pension money at stake. I think that everything we see from them and everything being thrown at Skype by want-to-be competitors is evidence that it is a genuine force with validity and staying power. The validity is best described along the grounds of saying that users simply love the product. You plug it in and it just works. The COOK Report on Internet May June 2005 It seems to be engendering a different kind of behavior and usage from people than does traditional telephony. I think at this point if you are out there trying to sell a VoIP product on the basis of it being simply cheaper PSTN quality voice that you are on the wrong track. COOK Report: I had no idea how true those statements are until I actually tired it, motivated in large part by David Reed’s post to my list. By the way what do you think of the May 19 2003 Viral Communications paper by Reed and Lippman? Enck: Fantastic paper. COOK Report: Do you agree then with the premises? Traditional Service Providers Blindly Want to Move PSTN Voice and TV to IP Networks Enck: Oh yes. Absolutely. I am very bearish about the prospects for the traditional service providers. I am a full 100% believer in the dumb pipe as their destiny. Given what is going on, I don’t see how they can avoid it. I think their track record of trying to reinvent and reposition their businesses so far has been absolutely dismal. I don’t think they possess the skill set to do the things they need to do. A lot of them seem to think they have the VoIP question already ironed out already. So the next thing they are going to do in Europe is charge down the path of recreating the television experience over broadband which I am not at all convinced that people necessarily want. I think those who say that a lot of what has gone on in the last couple of years with VoIP gateway manufacturers and border session controllers is an effort to recreate the PSTN in an IP world, when in fact Skype is saying: “your PSTN world is irrelevant we are going to give customers things they cannot get in the PSTN world.” They telcos are saying we are going to go out and bring you the old world but do it on demand. The only ones who may have it right so far are telecom Italia who have announced they are going to roll out a set top box with a DSL modem, a hard drive and a digital terrestrial TV tuner inside. They are making use of what is already there in Italy. They are not trying to reinvent the wheel which I think is very smart and is they way I have expected things to go. But I think a lot of these other guys are simply going to get fried. 65 Cable TV is a tough business. Why any telco operator wants to enter another world of pain is hard to understand. COOK Report: Agreed. I am coming to realize now that broadband is like a wellfertilized field on which you can grow and experiment and do an incredible amount of things. As long as you have an open field out there, the old closed loop networks can’t go anywhere unless government artificially extends their life. Enck: Exactly and what you can say in the case of the United States is that they have done so. I think Europe is turning into a very different picture. COOK Report: And Japan is certainly different. China and India will see broadband as a national resource not something to help old technologies maximize profits. Enck: Yes. But meanwhile in the US we will likely see a major collision between the interests of the municipal governments and those of the incumbents. In Amsterdam what the municipal government is trying to do is stimulate economic development and not give people cheaper web down loads. But meantime in the United States it seems like a great occasion to be a regulatory lobbyist. Symposium Discussion January 15 to February 8 2005 How VoIP Mixes with Wireless, the Enterprise and Other Markets Highlights VoIP Peering Architectures - Hard Installations or More Flexible Software Glue? Raj Sharma: It may be useful to think of IP telephony in the context of peering: (i) carrier to carrier peering, (ii) carrier to enterprise peering, (iii) carrier to consumer peering. Even in case of Skype, there is VoIP peering, specifically with its SkypeOut service, where Skype hands off traffic to another VoIP carrier to terminate the call to a traditional “‘black” phone. But let’s say that all 6 billion people on the planet are using VoIP and they have all discarded their traditional “black” phones, does the need for peering go away? I will submit to you that peering is required no matter what - it is required every time two networks which have different ownerships connect with each other. Along with peering come all the other peering agreements and arrangements such as topology hiding, security, admission and denial, settlements etc., etc. Forster: Can you expand on this? What value do these things deliver to end users? What percent of the calls must such a device be in the data path for the VoIP data packets? What’s the typical CAPEX per subscriber for these functions? I can see the need in a few cases in the data plane where there is dissimilar technology (e.g. no common CODEC in the end points), and a few more in the control plane (translation between different names spaces (Skype <-> E.164, etc.) such that there is no direct interoperation. Stastny: Even for that, peering will be necessary, it will also be done by the end-devices If you miss a video-codec, you just download it now already. I am using now the new beta of the Pulver communicator providing me with both SIP and Skype capability and it interworks with all IM versions. I am just waiting for a PDA version and then I have all my communications e. g. on a Qtek 9090. Forster: I have trouble seeing these peering functions as long-term important functions because the technology differences tend to reduce over time due to the network effect, rather than stay strongly different, or worse, continual emergence of new variants. In the past national interest in protecting local manufacturers was one of the forces that preserved silly differences, such as T-1 vs. E-1 framing, but I don’t see that happening now. Sharma: Now, will we be worried about the same agreements and arrangements that people have worried about in the old traditional TDM world - some may be, some may be not. However, a lot of this has nothing to do with technology necessarily. Peering agreements and arrangements are normal business constructs that various entities put in place as the normal course of doing business. Forster: I think an implicit assumption in what you are saying is that the business models can support the expense of negotiating and executing these agreements and deploying equipment required to support these agreements, whereas the alternative is to design the infrastructure and end points such that the need for these is minimized. In the past the cost of all this was small compared to the rest of the business, but IP telephony can drive down the costs of operation enormously and it’s especially noticeable as the voice service has been largely separated from the access network service -- a typical subscriber pays $X for broadband Internet access, and then $Y for a voice service, and Y<X. In some cases with certain restrictions Y=0, but in any case the separation of access from voice 66 service highlights the competition for the voice service. We’ve argued about various aspects of Skype, but I think no one will argue that what they’ve built is incredibly efficient -- very, very few people required per thousand customers. It just operates, with basically no operations staff. So the challenge for these peering functions is to justify their expense. Schulzrinne: The question is whether VoIP requires special application-specific peering, beyond the already-occurring IP-level peering at BGP level. Nobody is arguing with the need for the latter. I’m certainly not enamored with VoIP-specific peering - we don’t do email peering or web peering, either. COOK Report: I think Nextone has a window of opportunity - (space of time before the iceberg closes in). The Question is whether the window is six months or two to three years? And another thought occurs - if enterprises are having trouble making their VoIP systems talk to each, it would seem there is a business opportunity for someone to develop devices to make what would not otherwise interoperate do so. So you form a company, develop product and services. For some number of months you have a viable business and then OOPS - you aren’t viable anymore so you disband. Hopefully for a while you kept some people employed and performed an important service. And who knows - something weird happens and you may be viable far longer than you thought. You would argue better to do it right in the first place - fair enough. In this globalized economy can anyone even begin to tell what the best course is for the various segments of IT and Telecom? We each have our values and according to different values we come up The COOK Report on Internet May June 2005 with different answers. What then do IT and Telecom decision makers and investors need to know? I would guess they need to appreciate the complexity and understand how much they don’t know. Coluccio: I’m having some issues, probably internal ones, mostly, that I’ll have to work out myself at some point, but some also boil over into the real world, as well, that I’d like to air. Email hasn’t always been free of requiring peering fixes. I view VoIP’s placement on the Internet continuum today where email might have been during the early Eighties, when indeed it did require a host (no pun) of gateway functions in order for it to hop from one domain to another. And this preceded commercial exploitation. This didn’t stop email from becoming the most used of all Internet applications, and it won’t stop VoIP, either. What are we saying about the need to bridge between dissimilar voice protocols if we simply ignore the matter because it is unworthy, by whatever measure, of pursuit? What happened to the notion once voiced here that if you can get your message across to the party with whom you hope to communicate by slamming railroad Morse with a hammer and anvil, then it’s “good enough” if that’s all you’ve got at the moment to do the job? Would citizen Josephine be better served if she is forced to wait for the real thing, like Internet users in depressed areas who are waiting for the RBOCs to eventually “get to their neighborhoods” before receiving it, they get their first taste of broadband? Enduring the ugliness of kludge over IP is undoubtedly a better solution than no solution at all, as demonstrated by its ability to have already wreaked the effects of cannibalization on the carrier industry, even prior to its achieving singularity of form. Voice inter-working remains one of the few money making applications on the ‘Net. Even when it is no longer such, I think it is a topic that will deserve more attention than I’ve read here today. Reed: Frank - I agree with you that there will always be connectivity gaps to be filled in by whatever means will work. And often there are nice businesses solving such problems at low cost. When there’s a river in the way, getting people and goods across the river affords lots of opportunities for ferries, submarines, bridge builders, tunnel builders, airplanes, missiles, and dirigibles. But “peering” implies a single answer, and often it comes with the idea that the first solution will capture rents forever. In particular, peering is associated in my mind with interconnection of *exclusive* franchises, whether they are “corporate” or “municipal” or ILECs. With the idea that the government or management will see it your way and sign a long-term deal that locks out other solutions. In contrast, connectivity gaps can be filled by “a few lines of code” (well, maybe a few hundred thousand) running in a commodity blade server or a commodity router. And that hardware, and most of the software, can be picked up and reused for something entirely different. There’s a huge difference between such general purpose gap filling, bridging and protocol translation between legacy or competing service architectures (designed to be temporary until the industry sorts that particular problem out in a more elegant way), and a sustainable high-growth business. I’m in agreement with you that there is always need for a gasket or some filler where things come together; just suggesting that it’s important not to confuse terminology that sounds very architecturally “hard” like “peering point” with soft concepts like translation and glue. Making too much of the problem is just another version of George Gilder’s fantasy concept that 48V DC power will always be the optimal way to power communications gear, so Emerson Electric is a certain winner. (Ever wonder why 5.0000 V DC is the standard for TTL logic? I can tell you that it has nothing to do with physics or chemistry or maxi67 mum speed or minimum energy; neither is 48V anything more than a way to slow the introduction of commodity computer industry gear into the racks of telcos). Coluccio: Thanks, David. By the way, my use of the term “peering” was in deference to its prior use upstream by someone else, and I fully appreciate your clarification as to its more generally accepted use. I would use the term gateway or shim, depending on the problem, so your point is well taken. Speaking of usage behind the term “peering,” I just received this release from Converge Digest: ---snip: InfiniRoute Launches VoIP Peering Service for Wireless Carriers InfiniRoute Networks began offering a VoIP peering service for Wireless Carriers. The company, which has been offering a carrier-neutral Managed VoIP Peering (MVP) service since last year, now provides mobile carriers with the ability to terminate and receive international traffic using VoIP. http://www. infiniroute.com InfiniRoute Networks Inc. was established in 2004 as the result of the merger between two private companies, IP Deliver and Proficient Networks. Established in 2001, Proficient Networks provided a platform that measures endto-end IP network performance, responds quickly to automatically fix problems and prevents network congestion. Established in 2002, IP Deliver provided comprehensive VoIP network management services. IP Deliver’s Global VoIP connectivity product will continue to be offered as part of the company’s overall solution. Sharma: Interestingly enough InfiniRoute uses NexTone to offer VoIP peering service. But let’s say that “VoIP peering” is not the right terminology because it has been used in the context of IP peering, so let’s just call it some sort of “boundary function.” And even when there is no need to do any kind of VoIP COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA believe there may have been a connection between the two somewhere along the line. New Yorkʼs Broadband Gap Coluccio: All, I’ve often remarked about how unusual it was that there’s nary a hint of fiber to the home or node activity taking place anywhere in NY State, except, possibly, for the nodes that some ILECs feed for DSL services. Except, maybe, somewhere in state Verizon may have recently earmarked a location for its FTTP deployment, but I’m not sure where that is. After following this space since the beginning of time I don’t ever recall seeing mention of a fiber or extensive wireless build taking place in the state. In fact, if one looks at a plot of the two hundred or so FTTx deployments across the country, most of those are squarely placed in the middle of the country’s heartland. This point is driven home especially well in the December 2004 issue of Broadband Properties Magazine, which has done a rather complete survey of incumbent operator- municipality- and independent company- rollouts, to date. Today I was sent the URL below by the principal of a startup cable TV company, and from the looks of the report that it points to (“New York’s Broadband Gap”), New York City, itself, could use some additional silica and wireless coverage in quite a few of its neighborhoods and industrial park locations. I’ve not read the entire report yet, and probably won’t until tomorrow, but I thought I’d pass it along to the list at this time, just the same. Note that all of the neighborhood locations cited in this report are the same as those that were mentioned in the Daily News article that I posted here two or three days ago, where Ron Sege’s Tropos company was mentioned, which leads me to protocol conversion, a number of networks are getting deployed with a VoIPspecific Boundary Element. So even if this Boundary Element is not performing “VoIP Peering,” it is providing useful functions such as call admission control (the e-mail analogy is spam control), topology hiding, efficient call routing and CALEA. As long as VoIP is not a ”free” application like e-mail, there will be a need to do Least Call Routing which is one of the things that a Boundary Element does. If and when voice becomes free like e-mail, the Boundary Element is likely to participate in some form of efficient call routing given the real-time requirements of VoIP. Whether and how long the Boundary Element functionality stays stand-alone or becomes part of the router is up for debate. Matson: Think about the truck starting out on 5th Avenue in NYC making its way to Boylston Street in Boston - what “New York’s Broadband Gap” Prepared by Center for an Urban Future December 2004 http://www.nycfuture.org/images_pdfs/pdfs/telecom.final.pdf Who is CUF? From the back page of the report: snip: The Center for an Urban Future is a New York City-based think tank that fuses journalistic reporting techniques with traditional policy analysis to produce in-depth reports and workable policy solutions on the critical issues facing cities. For more information, visit our website, www.nycfuture.org. To sign up for our monthly e-update, contact cuf@nycfuture. org or 212-479-3341. This report was written by Jonathan Bowles with Tara Colton, edited by David Jason Fischer and designed by Julia Reich. Special research assistance by Nick Johnson.We also acknowledge the helpful contributions be received from many people. This report was funded by the Alfred P. Sloan Foundation. Additional program support was provided by the Bernard F. and Alva B. Gimbel Foundation, the New York Community Trust, the Taconic Foundation and the Rockefeller Foundation. The Center for an Urban Future is a project of City Futures Inc. City Futures Board of Directors: Andrew Reicher Mark Winston Griffith, Marc Jahr, David Lebenstein, Ira Rubenstein, John Siegal, Karen Trella and Peter Williams. are the peering arrangements between the two street locations? I have yet to see how in the case of IP telephony, there is any fundamental difference. Now networks (i.e. cities) don’t require peering arrangements - all the peering (if that’s the appropriate word) is in the hands/ head of the traffic itself. What am I missing? (Other than an ”operator”!) you been able to demonstrate the concept really scales? How Real Are Viral Radios? At the Media Lab, we are inventing ways to build systems that scale by cooperative adaptation to the propagation environment, that need no infrastructure. Retzer: I read the paper and did a little looking around on the viral network/ radios concept. Sounds great. How successful have your demonstrations been to date? Have you proven that the concept scales? We were working on CDMAtype concepts in the Pentagon in the 80s. As you begin to think it through you realize that bandwidth is theoretically infinite, however the net is limited by processing capacity of each node. Have 68 Reed: Jere - the truly simple and cheap viral radio network is a research agenda, not a business plan, so I’m very hesitant to predict deliverables. It’s important not to confuse a clear view with a short distance. So what have we invented? Well, I’ve been involved in filing 3 patents so far on the viral radio ideas, one on an invention done before I joined the Media Lab, and two after, and a couple more are being prepared. Not trying to be coy, but I can’t disclose the details because of the way the Media Lab sponsor contracts work - sponsors get first looks and first The COOK Report on Internet May June 2005 chance to negotiate rights. (It’s a good deal to join a Media Lab consortium or the Communications Futures Program!) The work involves both techniques based on advanced software-defined radios that have highly adaptable front-ends and fancy DSPs, and techniques that can be applied with mass-produced cheap radios of the kind that cost $10 per unit today the former could be just as cheap, but the volumes haven’t been there to drive the learning curves, though the new GnuRadio Universal Software Radio Peripheral is the next step in cost reduction on that paradigm (first radios were $20,000 per unit, USRPs under $500 per unit, ...). We also have a small NSF contract, exploring the use of networks of software radios to measure and adapt repeaters to indoor and dense urban propagation as it changes. A couple of students have written and published papers describing different approaches for cooperative physical layer repeating. Another student has been developing the concepts and techniques associated with “viral broadband” - to demonstrate his hypothesis that cross-connects within towns and neighborhoods and distribution of content into cheap local storage provided by the users provides a richer set of services that includes all “broadband”, but in a much more incremental and organic economic process than traditional command and control central service providers. But as I said, this is all early stuff. We aren’t the only people exploring the technologies that will lead to “viral” infrastructures. But I think our long-term vision is more clear - most of the other folks are trying to adapt ideas like multiantenna systems, MIMO, ... into existing legacy networks. Nothing wrong with that, in fact it’s hugely synergistic. Menard: While I too have a great deal of appreciation for what the MIT Media Lab is placing on this planet, including open sourceware and MIT Rooftop, the prospect of TV-Band 802.22-based viral radios does seem about 3-5 years out. By then we’ll all be well used to 26 mbps over copper... can you do 26 mbps over the air? Strange Concepts of Free Markets in the Muni Network World Coluccio; A point for Ron Sege: I’d imagine that the report referenced above should serve as a made-for-the-purpose road map to opportunities for your organization. What did you think about it? Has anyone else taken the time to go through it? I’d be interested in hearing of comparable circumstances in other large inner-cities, where I suspect similar conditions exist. Once again we see evidence of the principles we’ve discussed here in the past with regard to a potential reverse digital divide occurring in our day. It is typified by large cities receiving early generation implementations of low-speed ADSL, only to see, finally, after years of neglect, rural areas and smaller towns starting to step up to light speed rates and reinforced wireless platforms on their own, or through the late but better than never build outs by incumbent providers. I’ve been asked by someone off-list the following question, and was hoping you could fill the gap: “Does anyone have [a collection of] links to incumbent position papers outlining in detail what they don’t like about municipal networks and why?” It would seem obvious, but one never knows what could shake out of such a collection. In a similar way someone else offered me the following argument on the Gilder list about why governments should not get involved with networking, even after local elections and referenda by the people have endorsed them. It was in response to the article below concerning North Kansas City’s struggle with Time Warner: 69 Time Warner Sues NKC http://makeashorterlink.com/?J3DE1214A The reply (which is hardly essential reading, but interesting for the logic that it offers. I’ve found this and similar inclinations to be representative of the kind of thinking that exists in many parts, especially from those who have experienced the benefits derived by having a broadband connection in the home, oddly enough. Funny how that works, eh?): ---begin reply: Hi Frank---The Law Of Unintended Consequences has yet to rear its ugly head. For example, the purchasing power of groups of municipalities who may choose to create buying consortia could yield disproportionate influence on the choices of standards and suppliers. Not all of these choices will be identical to those that would have been yielded by a free market. Much capital might be wasted along the way, and superior technological solutions might be abandoned. Additionally, we run the risk of turning our brightest CEOs, venture capitalists, and visionaries into glorified lobbyists. The social cost of such a transformation would be enormous and might far outweigh any benefits we get from “jump-starting” the buildout. Those are the kinds of outcomes that keep laissez-faire ideologues awake at night (OK, not literally). Read “Empires Of Light” by Jill Jonnes for the fascinating story of how even a freewheeling capitalist like Edison was motivated to use the political process rather than the market in order to build public support for direct current, his own vastly inferior distribution technology. George Westinghouse and AC ultimately prevailed, but not before an expensive, drawn-out political and PR battle involving the exploitation of the horrors of the electric chair that played more upon the public’s emotional response to capital punishment than upon market considerations. You may choose to view this cautionary tale with its happy ending as proof that even politics won’t stand in the way of progress; I read it as a narrowly dodged bullet involving gobs of unnecessarily wasted time and capital. […] COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA ---end reply. have at least one example of a rural city however, the RBOCs’ fiber initiatives are that wanted a fiber POP but the ILEC the only sane way for them to proceed Baller: Actually, it’s really easy to jump would not hear of it so they managed if they want to stay in the wireline busiinto the industry position-paper loop. All to get a competitor to build one - which ness, because the total cost of ownership you have to do is read the latest industry instantly encouraged the incumbent to (TCO) of building and maintaining an “study,” and it will cite all the prior ones build one as well. Comments? Do we re- optically provisioned network pales in as “authority.” It get’s really amusing ally have more than anecdotal evidence the face of its far more expensive, over when you know when something flat-out of municipal builds discouraging private time, copper counterpart. And to do nothwrong was introduced and you then see it investment? What is “bad” versus “good ing at all is not an alternative for them, competition”? either, since the burdens and limitations quoted as gospel ever afterward. of a copper only infrastructure today just For example, a prominent attorney for Coluccio: Jere, I find myself interested won’t cut it. Therefore, the only soluthe cable industry once said in a confer- by what you stated, although I am not tion left for them to take - if they want ence handout that the municipal utility of entirely sure who the actors are that you to remain wireline players - is to install Scottsboro, Alabama, had gone to court refer to in your first sentence. When I put fiber, anyway. to stop its competitors from charging on my stereo viewer on, your comments competitive prices. In our oral presenta- are doubly intriguing to me because, The game of chicken you described, tions, I called the attorney’s hand on this, either way, whether one inserts “incum- where the incumbent wouldn’t budge noting that Scottsboro had not gone to bent” or “green fielder” in your first sen- until a competing entity gave them a court, but had filed papers with the FCC tence, it yields interesting and similarly good enough reason to, is now one of documenting that the incumbent was influential, yet different, results. legions. Were it not for Teleport’s move charging predatory prices over an exin 1984 to bring fiber to the businesstended period of time, and that the FCC Certainly, by the RBOCs taking the posi- es surrounding Wall Street, brokerages had confirmed in its annual report on the tion they have by beginning to expand there would probably still be using their status of competition in the cable indus- their reach of fiber into neighborhoods labyrinths of inter building pneumatic try that Scottsboro was right. Neverthe- and homes, competition on the part of systems to send their message missiles less, the attorney’s written presentation the cable cos goes up, accordingly, as to one another in order to execute trades took on a life of its own. The Scottsboro we’ve seen, along with everyone else’s. and settlements. canard was presented as fact in a “report” At the same time it has served to finally by the Beacon Hill Institute, and the legitimize in the industry’s, as well as Retzer: Why surprised? It seems to be Heartland Institute has now jumped on the general public’s, eye the matter of a recurrent primary argument by some the bandwagon. efficacy resulting from the spending of ILECs that they couldn’t possibly jusmoney on fiber builds. Thus, this tacit tify fielding system upgrades if they I’m puzzled by the arguments quoted form of endorsement has re-opened the had competition, especially from any above from the Gilder list -- Why is it pages of business plans that have been “government” entity. Isn’t that the raOK for the Bells to have a joint RFP collecting dust for eons, while the stone tionale behind their argument in PA for for FTTP but wrong for municipalities walling and sometimes hijinks over what example? to join in buying consortia, particularly was best to do had been disputed on an consortia such as NCTC that include almost universal level with fiber always It is interesting to flip the problem over scores of small private sector entities? previously receiving the short end of the and view it from the green-fielder’s viewHow does Edison’s (i.e., the private pipe, being regarded as nice to have, but point. If I’m right and competition is the sector’s) alleged abuse of the political still a blue sky future that was beyond the surest way to spur incumbents to action, but this places greenfielders at risk then it process prove that municipalities should short reach of copper. ;) may be that the surest method to get the be barred from providing communications services and infrastructure? While I’m at it, I think it needs to be incumbents to upgrade is for the public to said that the dollar numbers they’ve used compete with them. Retzer: Contrary to the notion that historically to represent the costs of fiber FTTH deployed equals competition that builds have been artificially inflated over Baller: [to answer Jere Retzer’s quesdiscourages providers, I contend that time to give them the breathing room tion] Here are two excerpts from the deploying fiber may be the best way to they’ve needed until now, and now, pro- FCC’s decision in the Missouri case, encourage the incumbents to step up to ceeding only due to the emergence of which we challenged on other grounds: the plate. Seems like there is nothing alternative sources of services in their like the prospect of losing a market to midst, as presented by cable operators From the unanimous Commission opinmove comfortable incumbents out of and wireless alternatives that heretofore ion (footnotes omitted): their comfort zone. How else could we have only been threats, but are today now “10. While the legal authorities that we explain the innovation and investment materializing as we speak. must look to in this case compel us to that began with Carterphone and the ATT breakup decisions? Within Oregon, we Viewed through a slightly different angle, deny the Missouri Municipals’ petition, 70 The COOK Report on Internet May June 2005 we reiterate the Commission’s urging in the Texas Preemption Order that states refrain from enacting absolute prohibitions on the ability of municipal entities to provide telecommunications service. The Commission has found that municipally owned utilities and other utilities have the potential to become major competitors in the telecommunications industry. In particular, we believe that the entry of municipally owned utilities can further the goal of the 1996 Act to bring the benefits of competition to all Americans, particularly those who live in small or rural communities. We emphasized this fact in our August 2000 report on the deployment of advanced services. In that report, we presented a case study detailing advanced services deployment in Muscatine, Iowa where the municipal utility competes with other carriers to provide advanced services to residential customers. “We noted that the degree of advanced services deployment in Muscatine, which has three facilities based, high speed service providers for residential customers, including the municipal utility, is due in part to Iowa’s legal environment, which has encouraged municipal involvement in the deployment of advanced telecommunications services. Our case study is consistent with APPA’s statements in the record here that municipally owned utilities are well positioned to compete in rural areas, particularly for advanced telecommunications services, because they have facilities in place now that can support the provision of voice, video, and data services either by the utilities, themselves, or by other providers that can lease the facilities. We are also encouraged by the comments of Missouri River, which states that it is comprised of municipally owned utilities that serve communities with populations of less than five thousand people in Iowa, Minnesota, North Dakota and South Dakota, and that its members have installed fiber optic facilities that they could use to provide telecommunications services in markets where there are currently no competitive alternatives.” From the separate statement of Commissioner Ness (footnotes omitted): “In the Telecommunications Act, Congress recognized the competitive potential of utilities and, in section 253, sought to prevent complete prohibitions on utility entry into telecommunications. The courts have concluded, however, that section 253 is not sufficiently clear to permit interference with the relationship between a state and its political subdivisions. “Nevertheless, municipal utilities can serve as key players in the effort to bring competition to communities across the country, especially those in rural areas. In our recent report on the deployment of advanced telecommunications services, we examined Muscatine, Iowa, a town in which the municipal utility was the first to deploy broadband facilities to residential consumers. The telephone and cable companies in Muscatine responded to this competition by deploying their own high speed services, thereby offering consumers a choice of three broadband providers. It is unfortunate that consumers in Missouri will not benefit from the additional competition that their neighbors to the north enjoy. “I urge states to adopt less restrictive measures, such as separation or nondiscrimination requirements, to protect utility ratepayers or address any perceived unfair competitive advantages. Allowing the competitive marketplace to work will facilitate the type of innovation and investment envisioned by Congress when it enacted the Telecommunications Act. I join with Chairman Kennard and Commissioner Tristani in urging Congress to clarify its intention in section 253 with respect to prohibitions on entry by municipal utilities.” Coluccio: Jim, thanks for both replies to my request on the matter of incumbent position papers. The second one concerning the MO case was quite illuminating. And your example of the Joint RFPs for DSL and FTTP caused me to wonder why I didn’t think of that. It fits superbly. I suppose what I was in search of is the stance that the incumbents are taking as they approach their lawyers and lobbyists with instructions that include justifications behind their stated positions. “Here’s what we see the issues being, what our position is as to why we think 71 there is an unfair playing field being imposed upon us, the near term sacrifices being worth the long time wait, why cutting back on staff is essential and therefore the time to repairs on weekends are naturally going to be excessive.” That sort of thing, if such exists. But in retrospect, I suppose that is all held very close to the vest or available only in oral form, mouth to ear, but thanks for the replies just the same. Google Wants Dark Fiber On January 17 Coluccio: Take a look at: http://news.com.com/Google+wan ts+dark+fiber/2100-1034_3-5537392. html?tag=sas.email Job listing spotlights vague plan to develop global fiber backbone. But why? ”Google wants ‘dark fiber’” By Evan Hansen “Is Google planning to build a global fiber-optic network from scratch? And, if so, why? “ “The question has cropped up in light of a recent job posting on the search engine giant’s Web site seeking experts in the field.” “Google is looking for Strategic Negotiator candidates with experience in...(i)dentification, selection, and negotiation of dark fiber contracts both in metropolitan areas and over long distances as part of development of a global backbone network,” the posting reads in part.” [Snip] Peter Cohen: Couple of thoughts here. Google’s strategy and deployment is such that they are interested in reaching as many destinations, both corporate for customer base and network for peering as possible. There are looser networks that will peer with Google despite ratios, given their reach, or saving costs on their upstream to reach Google (win-win situation). There are others that have stated such policies as you need your own backbone, customer base, etc... in various COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA places and irrelevant of ratios, Google may get some of those by expansion in particular areas or countries. The other option is that Google has some grand scheme that this metro net is needed for in deploying a new product or service. Could be something like Akamai for distribution of content that is cached, cachable, Ca$hable!. Schulzrinne: How about the simple, but far less fun, motivation that they might want to interconnect their various data centers more cheaply? Coluccio: That could be taken in a way similar to how municipalities start out by wiring their muni buildings, first, and then gradually pursue, through a series of phased network extensions, enterprises and residences later on. Or it could be nothing at all, as I remind myself that this “tip” came from a bulletin board from a Light Reading subscriber. Although it’s entirely logical that, like other enterprises who can justify a critical mass, Googie should seek to optimize on their backbone costs, as well. Giving it further thought, though, after witnessing their rolling out storage and mail offerings over the past year, it’s clear that this company is aspiring to more than a librarian’s status on the ‘Net, so the prospect that they’d be seeking direct plumbing to NAPs and to larger enterprise institutions, directly (if you extrapolate) for whatever reasons, at first, is not far fetched. Stop and consider for a moment of the negative growth impact - as measured by shift and lost future opportunities - this would have on a bean counter’s account of the volume of traffic traversing the ‘Net versus those of private shunts at a lower layer in the stack, once an application such as Google’s and its nowancillary but potentially future-dominant services from, a bandwidth consumption perspective, have shifted to a “customerowned” or a “private backbone” framework. Retzer: I find that hard to swallow. Google benefits as much or more than any company from a full interconnected network. Just because they are thinking about acquiring their own high-speed links in no way implies they would remove access to their content and services. Consider that Google is continuously inventing new services and that their popularity stems at least in part from the high speed as well as quality of response they give users and it is small wonder they are interested in fiber. Google operates in several languages and countries now - the Internet is far less US-centric than it was. One service alone, offering GB e-mail storage for GMail may be sufficient to justify the links. Coluccio: With respect to the prospect of a company like Google’s building their own infrastructure and taking a slice of IP traffic off the open Internet along with it, Jere Retzer wrote: “I find that hard to swallow. Google benefits as much or more than any company from a full interconnected network. Just because they are thinking about acquiring their own high-speed links in no way implies they would remove access to their content and services.” Gill: I agree. Coluccio: True, but that is not what I was getting at. Google itself or any other like it would always have a desire to extend their content and services to users, whether those users are large enterprises or consumers, but such does not *always* require what has now become regarded as access through one-on-one connections with them. Take for example how large firms now access carrier hotels and collocation sites on their own, by virtue of lambda directs they execute through their own dark fibers, which they rent from fibercos or build on their own. Can you envisage a Google cutting deals with the Merrill Lynches and GMs of the world, heck, with any enterprise for that matter, whereby Layer 1/2 piping to their data centers and communications center avails to end users the same content and services as those companies’ users would otherwise have to search out over the Web? Think of it as a drop shipment in bulk capacity form (daily uploads to 72 client caches?), if you will. A collateral benefit of this could even be seen in the heightening of security, too, as well as offloading of traffic from main access router ports to the Web, where business as usual is conducted. Gill: The problem is that ports cost money. For a few tens of critical suppliers, sure. However, it doesn’t scale to several thousand such. Building and maintaining a fiber infrastructure that homeruns to the customer base costs money. You want to get the benefit of amortizing your OPEX and CAPEX across as many connections as possible over the same pipe. Coluccio: Looking forward, I believe we’ll see more of this type of WWW bypass situation unfolding where it makes sense, which, if not done identically to what I’ve just described above, then in some similar form. Such capabilities will be enabled cost effectively through the inherent abilities and economics afforded by optical switching, when done for the purpose of creating quasi-permanent virtual links to SPs at Layer 1/2 from colocated network elements owned by end users in the same or colos, or even between virtual colocation centers. Gill: This looks good on paper. Places where rubber meets the road, not so much. With the rate of current trouble tickets on our pipes, there will be a full time department just managing and troubleshooting the network. This is not Google’s core functionality. Coluccio: There is always the other way out by using a gateway built for the purpose, as an alternative that would allow the accessing of the WWW directly, and then crawling over to the targeted Google or cache site, as is normally done. And while I view the eventualities I cited above as inevitable, I’m only stating them as my interpretation of a trajectory that’s already in motion, and not from an ideological slant. I hope that that was a sufficient amount of CO2 to douse any nascent flames that might have been smoldering. ;) And if Google currently operates using several languages now for the purpose of The COOK Report on Internet May June 2005 mass customization, and seeks through further market targeting to entice additional business through a narrower approach to customized products, then think what having a captive usership on the other end of a dedicated pipe might portend for their marketing potential down the line, sans the overhead of much other peoples’ services and gunk. Gill: Or license their software and run it in the customer datacenter, instead of providing pipe. Cohen: Ooh. I forgot one thing. The guy in charge of infrastructure to some extent at Google was at Yahoo before that. He had plans there to build Yahoo a global network of some sort and probably took that plan with him. Gill: Precisely. There isn’t a quantum jump between “100% transit in one location” to “global backbone.” At some point, it makes more sense to gradually build out the backbone like a large ISP and run that between your major data centers, and to peering locations. Its a purely numbers play. AOL did something very similar to this, gradually moving from a few transit points to a global backbone with no transit. St. Arnaud: I suspect the reason that Google may want dark fiber is the same economic pressures that are driving many media companies to reduce Internet transit fees by locating services at major carrier hotels around the country. Internet transit pricing seriously penalizes organizations that have large asymmetric data flows such as film and multimedia companies. Equinix specializes in the business of hosting these companies servers at various major nodes throughout the US, so that these companies can directly connect to Tier 2 and Tier 3 ISPs and avoid transmit fees from Tier 1 ISPs. Bill Norton at Equinix has written several good papers showing the economic cost savings of this type of business model. Of course, these organizations still have to connect to these server hubs and this is where dark fiber comes in. Coluccio: Thanks, folks, for some great replies to my hypothesis and troublesome predictions. Google as an Example of When to Route Versus When to Switch Coluccio: I would hope that we keep in mind here that although the name Google presents itself, as the article noted, because of its mention on a Light Reading bulletin board, the same principle might apply to others, and not necessarily even to Google, itself. But it was interesting to read some posts that suggested it might actually be them. The Equinix model, if inclusive of end user organizations on an equal footing with SPs and content providers, would allow for much the same effect as I outlined upstream. Dumb Question: Does Equinix currently treat user organizations on an equal footing with service providers, when the user pipes itself to one of the Equinix centers? I’ve lost track, but at one time, I believe, that they did not, since they claimed it would compromise their allegiance to their best customers’ needs. Now to look at it a little differently - and with regard to a large content provider doing what I suggested in my previous post on this subject, Vijay stated: “The problem is that ports cost money. For a few tens of critical suppliers, sure. However, it doesn’t scale to several thousand such.” Frank’s reply: Good point, except I qualified this by stating that handoffs and peering would likely take place in colos and carrier hotels, where large enterprises now make their presence known either directly or by virtue of lambda hops made possible over an AboveNet or L3, or others’ interconnected lambdas that are split off at the customer site and beamed to the colo in question over special arrangements, Which aren’t so special anymore, and becoming commonplace for the very largest users. 73 The fiber company, in turn, has fiber routes going to where it needs to be to satisfy the needs that I cited. Where this was once the sole province of Fortune 50s in the past, it is easily attainable by smaller companies who have dark fiber access in their fabric, and will be commonplace for the 1000s, soon, who are situated within reach, not to mention those of educational and research entities. It’s interesting to consider the tradeoffs. Colo and carrier hotel distances today are accommodated with short- and very short reach, or VSR, optics, or even over copper patches using Cat 6 for GigE and soon even 10GigE speeds at reduced distances, in some qualifying cases, Each of these options costs far less than the ports associated with the rigmarole of provisioning additional physical access plumbing at the customers’ snail locations and their attendant monthly recurring costs for rental. And then there is always the Ethernet peering potential via the electrical interface that can be had at the colo or meet point, if optics become an issue either due to distance-port-cost tradeoffs, or availability. From a practical and imminent perspective, you’re probably right for the larger number of even very big enterprises, though, at least at the present time. But I tend to look out 18 to 24 months and in that case what I’ve stated is not far fetched from a trending viewpoint. I appreciate your input and those of others on this topic very much, nonetheless. Thanks. Earlier Bill St. Arnaud stated: “Of course, these organizations still have to connect to these server hubs and this is where dark fiber comes in.” To expand on a related one of your points, Bill, and to pick up where I left off with Vijay about enterprises seeking similar peering and transit capabilities although not exactly the same - as those of the ISPs and BBs: I may have mentioned this account here COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA before, or on some other list, not sure. If it looks too familiar I won’t be offended to hear “next” being clicked throughout the ether. A large client of mine signed onto a large parcel of property that was out of state about five years ago, and I was commissioned to do the outline for its macro architectural statement for what amounts to being a campus design. This included the base buildings to the WAN and ISP worlds, and most of the innards as they related to a very large vendor’s then socalled “New World” architecture. From there, others were assigned their appropriate pieces and life went on. At an early stage brainstorming session I suggested that, in addition to the other five dominant facilities-based providers’ fiber routes, the client should attach to a dominant colocation site in NY City for the purpose of meeting service providers at the colo’s “meet me” points to become a part of its “bandwidth pool.” I made the mistake at that time of calling it a bandwidth pooling point. Recall the expression, bandwidth pooling point? Anyway, nary a single citizen in the room knew what I was talking about, and with fifty people sitting around the horseshoe all looking you in the eye at the same time, that can be a very lonely feeling. After some discussion, which took a couple of weeks in conveying the concept adequately, the concept became clear, but the defaults had already been set in motion, which you can read as, no one wanted to be confused with facts at that point in time. As the project was nearing completion in 2003 I learned that MFN had just attached this client to the same colo I had suggested three years earlier. They also built out about 8,000 sq. ft. of cage area for both production and development server farms under the same roof.. This has saved them over the past year or so très beaucoup de dollars, due to the obvious improved methods of interconnect allowed by close-proximity handoffs to its service providers and to other locations within their own customer-owned fiber optic based MAN. Previously, the client was using six proprietary, interlocking OC-48 SONET rings to achieve a fraction of what they are capable of carrying today, at a much higher cost, not to mention the added aggravation of managing an increasingly IP world through SONET orifices. To a large extent the existence of ILEC central offices, for this customer, at least, became a superfluous proposition for the purpose of reaching their highest capacity needs, if not potentially detrimental to their overall mission. So, as you can see, even when this form of solution was advised to the very largest of banks only three years ago, it was shunned. The evolutionary path for them to awaken to the reality of the approach was 24 months. So, I guess I should have waited the 18 to 24 months I spoke of earlier, and then I wouldn’t have felt so alone in that u-shaped conference room, way back when. Google VoIP Stastny: Another idea why Google may want dark fiber is on Tom Keatings Blog http://blog.tmcnet.com/blog/tom-keating/VoIP/VoIP-blog/google-VoIP.asp Schulzrinne: Calling this idle speculation would give it too much credibility. Unless they plan to supply their yellow page business customers with fiber or run fiber to residences, none of these VoIP calls would ever touch their network. They’d only make money if they became another run-of-the-mill Tier-1 ISP, of which there is not exactly a shortage. Some of these have gone through the Chapter 11 rinse cycle, so their former stock holders (and now wallpaper holders) have paid for their capital expenses. Stastny: Of course Henning is correct that VoIP calls from the end-user clicking on the link, but on one hand they may want to connect the other side. On the other hand the idea of Google as ISP may not be too far-fetched: Think about YahooBB in Japan and Fastweb in Italy (providing FTTH in all bigger Italian cities between 50 to 100 Euro/ month). The 100 Euro provides you with at least 10MB and a set-top box containing a video-cam for real-time video communications from home. Italians like to participate in TV-shows interactively Wetzel: That’s funny, instead of having access providers that desperately want to provide content with the hope of making money with it, to compensate the loss they experience in selling low cost accesses, we may see content providers which will provide accesses to end users and which will make money with both. Coluccio: It sounds similar to, but not exactly, the evolution of the cable TV cartel. Oops... did I write ‘cartel’? Of course, I meant to write “industry.” Stastny: Tom Keating was right: see: http://www.timesonline.co.uk/article/0,,2-1454225,00.html Citation: Mr Hewitt said that a Google telephone service could be made to link with the Google search engine, which already conducts half of all internet inquiries made around the world. A surfer looking for a clothes retailer could simply find the web site and click on the screen to speak to the shop. See the following quote: Google gears up for a free-phone challenge to BT by Elizabeth Judge, Telecoms Correspondent http://images.thetimes.co.uk/images/trans.gif GOOGLE revolutionized the internet. Now it is hoping to do the same with our phones. The company behind the US-based Internet search engine looks set to launch a free telephone service that links users via a broadband Internet connection using a headset and home computer. The technology that will enable Google to move in on the market has been around for some time. Software by the LondonContined p. 76 74 The COOK Report on Internet May June 2005 Millimeter Wireless Links services to service providers, financial institutions, enterprises and government agencies. It appears that it’s getting easier with time to establish wireless links that possess fiber like speeds, as Gigabeam has done with its 1Gbps millimeter radio system. The story below about this firm and Stealth Communications, a VoIP service provider, is a fairly impressive, if you think about it. GigaBeam anticipates installation of the first Wi-Fiber GigE link for Stealth in the current quarter, providing high speed connectivity of one Gigabit-per-second across the Hudson River between a Stealth customer’s data center in Jersey City, NJ and Stealth’s Manhattan access point at 60 Hudson St., New York, NY. One Gigabit-per-second is the equivalent of 647 T1 The ability to drop ship an antenna and its associated 1 Gbps lines or 1,000 DSL connections. millimeter-based radio system on a rooftop or window sill anywhere there is line of sight from a hub location, and then Lou Slaughter, Chairman and CEO of GigaBeam Corporation disrupt or otherwise supplant the local telephony operator, stated, “GigaBeam’s ultra high speed wireless fiber technology is now upon us. Terabeam once portended to be able to this is an ideal solution enabling Stealth to provide a high speed through the use of its free space optical systems at one time, connection between customer locations and Stealth network but for reasons not germane to this post, they didn’t. Giga- access points.” Until now, Stealth has relied on traditional fiber beam and others, however, appear ready to deploy in ways optic cables to connect its customers and Stealth locations that give the sense of being more secure, and probably are, together. during times of inclement weather, hence will likely succeed where others did not. I believe we’ll see many variations of With GigaBeam’s solution, Stealth customers will be able to the scenario painted below, both for providing dedicated ac- bypass local infrastructure and other right-of-way obstacles cess, and in an ancillary way in which high capacity systems that prevent quick and cost-effective high speed connections such as these, which will likely support more backhaul for using traditional physical fiber. “The initial link highlights hot spots and meshed nets, thus fostering ad hoc alternatives, GigaBeam’s significant advantages over traditional fiber for as well. The impact implied for data solutions should be- primary and redundant fiber speed connections,” Slaughter come obvious as you read through the release. Do you sup- added. pose that an ILEC might attempt to use the same approach at some point, in order to thwart encroachment from such GigaBeam’s Wi-Fiber technology is not only expected to restartups as these in the future? Dare I ask about a municipal- duce costs and provide additional levels of access redundancy, but it is also expected to improve network performance by reity, or a utility, the same? ducing network access latency by as much as 30% or more. The first link will be implemented in Stealth’s innovative Financial GigaBeam, Stealth Expand VoIP 01.17.05 Extranet which will enhance performance and reduce latency HERNDON, Va. -- GigaBeam Corporation (OTC Bulletin for the exchange of traffic among its members. Board: GGBM - message board), announces an agreement with Stealth Communications(R) for installation of its ultra Shrihari Pandit, CEO and founder of Stealth Communications said, “Using GigaBeam’s Wi-Fiber wireless fiber solution, we high speed Wi-Fiber(SM),(TM) wireless fiber solution. can now address the demands of a much broader market. Now, GigaBeam’s Wi-Fiber provides communications access with with a direct fiber-speed wireless Wi-Fiber Ethernet circuit, the capacity of fiber but with significantly less latency than our customers can connect to our unique portfolio of services, fiber to enhance Stealth’s unique communications services including Stealth’s Financial Extranet and The Voice Peering offerings. Stealth is New York City’s largest Internet gate- Fabric(TM). The GigaBeam-Stealth solution provides high way and offers Internet & Ethernet access, VoIP, and data performance, low cost access without the expense of running physical fiber, and allows bypass of local infrastructure.” SIP or Skype. Distributed has always been best. Peer–to-Peer SIP On January 21 Stastny: The following information was distributed via the sipping mailing list. Jennings: David Bryan and I have submitted an draft of some early ideas on P2P with SIP. Until it shows up in the repository, you can find it at http://p2psip.org/ Roberts: This SIP approach is how it should work! Hopefully the world will convert. It is a better technique than today’s 75 David Reed: Larry captured my reaction perfectly! This could be big if adoption becomes viral (but if it becomes yet another “what’s the business model” debate among incumbents saying “after you, no after you”, it will die). So the question is what edge-based marketplace will use it and become the driver? (Something cool and appealing, like Quake, could do it). COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA New Push for Wireless Mayors see the perfect political package: They get to address economic development, digital divide, public safety, and more Retzer: New push for wireless access by Qwest and others. broadband access and lower prices. All with one low-cost investment in infrastructure that non-incumbents are increasingly Front page of today’s WSJ. willing to fund (see below). See Philly Mayor Street’s address Alex Goldman: It proclaims “Dual Mode a Step Closer to this week to the National League of Cities. Reality “ Unbundled local loops keep getting more expensive so the “Companies are ironing out the final wrinkles in technology lesser-incumbents like Earthlink, AOL, Sprint, MCI are more anxious than ever to promote alternatives. that will allow cell phone subscribers to use IP networks.” http://www.isp-planet.com/fixed_wireless/technology/2005/ Verizon made a big tactical blunder by going after Philadelphia as they did. They simply called attention to the power of the new kineto_uma.html technology and galvanized more municipalities into action. My understanding is that Kineto is a relatively small company that could win big time through the UMA relationships it has with several major carriers. I’ll have a follow up article on Monday, about a conversation with Monica Paolini of http://www.senzafiliconsulting.com and her latest UMA research. Intel has been promoting WiMAX heavily as a fast, low cost, simple alternative last mile. Non-incumbents like the messages but since WiMAX is not available and metro Wi-Fi mesh is, they are using it now. I think this phenomenon has legs. Cable companies are looking By which I meant to say that today and this week seem to be at this, as are large retailers, both to pick up more customers and leverage large backbones they already have. Maybe Google, quite the wireless IP week. A trend. Why now? too?? Sege: Because: At least 125 cities in the US are doing it today and the technology (metro Wi-Fi mesh, etc.) works and benefits are becoming apparent. Chaska, MN now has 2,500 users out of 8,500 homes, paying $16/month for an average 1.5Mbps throughput. No net churn since billing began in the fall. based company, Skype, has been downloaded nearly 54 million times around the world but no large telecommunication firms have properly exploited it. BT, which connects seven out of ten British households, has developed its own Internet-telephone service. However, the telephone giant, which has the most to lose if the new technology takes off, has been reluctant to promote it heavily. Julian Hewitt, senior partner at Ovum, a telecoms consultancy, said: “From a telecoms perspective there is a big appeal in the fact that Google is a search operation - and of course the Google brand is a huge draw.” Mr Hewitt said that a Google telephone service could be made to link with the Google search engine, which already conducts half of all internet inquiries made around the world. A surfer looking for a clothes retailer could simply find Sandel: Hi Ron, Would you happen to have a pointer to Philly Mayor Street’s address this week to the National League of Cities? Sege: http://www.broadcasturban.net/webcast/uscm2005/mon_ session.html the web site and click on the screen to speak to the shop. The basic cost of making calls across the Internet is almost nil. The real cost is in developing the software; after that, the service exploits available Internet capacity. However, charging does become necessary to link internet calls with the traditional phone network. In addition, the sound quality of calls across the Internet can be poor and the connections can be less reliable. A recent job advert by Google’s on its website calls for a “strategic negotiator” to help the company to provide a “global backbone network” - a high-capacity international infrastructure. By investing in capacity, Google could circumvent the problems of quality and reliability and guarantee better service. 76 Although Google is reluctant to talk about its plans, the logical use of such a network would be to help to support a new telephone service. The company would buy capacity cheaply, by taking up slack capacity left behind when the internet bubble collapsed in 2001. Around the world, thousands of miles of fibre-optic cable remain unused because the amount of speculative development vastly exceeded demand. Such capacity would be available at rock-bottom prices today. Elsewhere in the world, using the internet to make phone calls has caught on more quickly. In Japan 10 per cent of households already use the so-called “voice over internet protocol” and an internet service offered by Softband has 4.4 million subscribers. Its growth has depressed revenues of the local telecom group, NTT. The COOK Report on Internet May June 2005 In the US, a company called Vonage offers customers unlimited calls each month for as little as $24 (less than £13). Big companies and multinationals that make huge numbers of long-distance calls are also increasingly switching to internet calls to try to slash their bills. Google, which was founded in 1996, built its business from scratch by offering a fast, reliable and free Internet search. It gradually transformed into a highly profitable company by offering commercial services, including sponsored web links. Its most up-to-date figures show that, in the first nine months of 2004, Google made a profit of $195 million on revenues of $2.1 billion. AT&T Looks Beyond “Number, Please” Coluccio: Interesting headline. I wonder which “number” the company will be referring to, by the time it transformed itself into a new being ;) AT&T Looks Beyond “Number, Please,” Ken Belson January 22, 2005 “For a century, AT&T was known as America’s premier phone company. If Hossein Eslambolchi has his way, that label will go the way of the dodo bird. As chief technology and chief information officer, Mr. Eslambolchi is the technological strategist behind AT&T’s ambitious turnaround plan to become a data transmission company selling an array of software products like network security systems - with phone calls being just one of many digital services. Indeed, for the first time, voice calls generated less than half of the revenue in AT&T’s corporate business group in 2004. A few years ago, this approach was heresy at AT&T, where connecting calls was the cornerstone of the former monopoly’s business. But with falling prices, growing competition and cheap new Internet phone services from start-up companies, AT&T’s future depends more than ever on vigorous cost-cutting and focusing on its worldwide data network. “There is a sense of urgency,” Mr. Eslambolchi said. “We have no other alternative. We have to do this to survive.” The strategic shift was evident in the company’s quarterly earnings. The company said Thursday that profits jumped 84 percent in the fourth quarter, to $625 million, beating Wall Street’s estimates, thanks primarily to cost-cutting efforts. AT&T eliminated 23 percent of its work force last year and has automated large parts of its operations. The cost cuts, however, only slowed the effects of a price war that drove the company’s revenue down 10.2 percent, to $7.3 billion, in the quarter. The worst is far from over. The company expects sales to decline another 15 percent this year. The way to stem the slide, Mr. Eslambolchi contends, is to merge the hundreds of computer systems AT&T created over the years. With phone calls and data now transmitted increasingly via high-speed data lines using Internet protocol, the need for multiple systems is also diminishing. By reducing the number of networks and systems, Mr. Eslambolchi has been able to reduce the size of AT&T’s once gargantuan work force and strip away layers of management. That is easier said than done in a company filled with internal fiefs and unaccustomed to responding quickly to market demands. Mr. Eslambolchi compared the company to the United Nations, where a multitude of managers have veto power. “The biggest challenge is not the technology,” he said, “but being able to change the culture.” [Big Snip] For all the turmoil that years of declining sales have caused, AT&T’s workers are just getting Mr. Eslambolchi’s message. Since 2002, he has dismantled 381 of the company’s 800 computer systems designed for specific products and ser77 vices; he said he expected to have just 20 by 2007. AT&T is also using more software to route more of its phone and Internet traffic. By getting rid of bulky circuit switches, the company is significantly reducing costs connected to operating old-fashioned switching stations. Mr. Eslambolchi is also pushing engineers in Bell Labs to develop software for computer firewalls and security systems that detect viruses days before they attack a corporate client’s servers. These kinds of advanced services grew 13 percent last year and now make up 10 percent of sales in AT&T’s business group. [snip] The Web site handles about 2.6 million transactions a month. Billing problems for business customers have fallen to 3 percent from 11 percent and the customer support center now handles 80 percent fewer calls. By automating so many tasks, the number of customer service and network jobs has plunged to 22,000, from 55,000 in 1999. In all, AT&T has reduced $2.5 billion in operating expenses since 2002. [snip] Verizon Communications and SBC Communications, in addition to aggressively pursuing AT&T’s remaining residential customers, are also aiming to lure AT&T’s business customers with discounted services. The Bells’ efforts on this front have quelled market speculation at least for now that one of them might acquire AT&T or another long-distance carrier. [snip] http://www.nytimes.com/2005/01/22/ business/22phone.html?ex=1107439672 &ei=1&en=8de63670cb94e4e8 Odlyzko: Well, Hossein Eslambolchi, the chap featured in the article, is famous (or notorious, depending on your point of view) for his “concept of one and COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA concept of zero.” We’ll see where AT&T ends up. The idea to get into services is an obvious and overdue one, the main question is whether there is enough time for AT&T to accomplish this. My guess was that this would work out only if AT&T, MCI, and Sprint were combined, to lessen some of the price-cutting that is driving them into the ground, but I don’t see anyone able to do it (and with the recent Spring-Nextel deal, it would be hard to arrange). Coluccio: It’s odd that you should mention Sprint-Nextel here, because AT&T is in the process, as we speak, of becoming a Mobile Virtual Network Operator (MVNO) taking partitions off of Sprint’s wireless platforms, effectively becoming a reseller of Sprint’s wireless services. When Seven-Eleven does this they are called a MVNO. And in my book, that is exactly what AT&T has committed to doing, as well. That sounds to me like they’re still trying to garner revenues from voice minutes. Doesn’t it sound that way to you? The company has always been good for making a splash far in advance of perceived threats. As when, Tom Evslin, then recently brought on by AT&T from Microsoft during the mid-nineties proclaimed that if anyone was going to cannibalize AT&T’s voice service revenues with IP Telephony (what was then called variously IP Phone, or simply I-Phone), then it would be AT&T itself that was going to do it, and no one else. They’ve had ten years to inflict said cannibalization on themselves. Have they succeeded yet? Odlyzko: Yes, AT&T is planning to become a MVNO, but this is largely in order to be able to offer a complete bundle of services to its customers. There is not much money in being a reseller of somebody else’s service in general. There is nothing wrong with trying to garner revenues from voice minutes. My argument has been for years that the wireless industry has been negligent in ignoring the opportunities in more wireless voice and higher quality voice. Once we get to real ubiquitous and inexpensive wireless broadband state, that opportunity will be gone (as it is pretty much gone in the wireline area), but right now it can still be exploited, but these guys are still mesmerized by content. AT&T has done very little cannibalization through IP Telephony. Coluccio: I can’t say I disagree with you on any single point, although I maintain a view that voice will be reduced along with email at some point, taking it out of the running as a means to enjoy appreciable revenue. At least as it applies to the consumer markets, if not a great swatch of enterprise, as well. It offers constantly diminishing rewards, in other words. And even if AT&T has not cannibalized “itself” in terms of IP telephony, yet, it has, nonetheless, joined the fray that points to the same end game by introducing and supporting its own flavor of parasitic VoIP - CallVantage offering which uses best effort bandwidth side by side with the MSO’s QoS-enabled VoIP offerings under the heading of PacketCable. What we’re seeing here, in effect, is a food chain that’s been turned on its head, where fish that aspire to smaller and smaller status continue to kill off (as opposed to eating) their larger encroachers, if that makes any sense at all. All of which leads me to ask: What is the probability that PacketCable and other forms of assured voice offerings will become profitable enterprises for those carriers and SPs offering them, while continuing to maintain the low price points they used when those services were initially introduced? Since their price point is at least in part based on a perceived penetration rate, I guess my question could have as easily read: Will they meet their projected penetration rates in the face of newer ‘Net-based alternatives, with the latter often times classified as “free”? I and many others like me have already taken to using Skype for a number of purposes for which I’d have used a second or third line, recently. And where I might have opted for a Lingo, or a Vonage or a CallVantage service a couple of weeks ago to replace my aux line now in the home <and now my cable operator’s 78 voice is becoming available> I’m now holding off to see if it’s worth my time to again negotiate the nuances of another new voice services provider. Retzer: On January 17 the Wall Street Journal announced - Verizon, Yahoo Plan Joint Broadband Service. Hoping the subject line will get Andrew to jump into the conversation. Interesting that Verizon in this article still seems to think content is king, along with Comcast. Andrew’s studies lead to the opposite conclusion. Odlyzko: Responding to your prod, I don’t see how this story necessarily proves that Verizon regards content as king. (They may very well do so, but this story does not prove it.) There is nothing wrong with content, nor in telecom service providers being interested in it. The issue is how much attention and resources should be devoted to it. As an analogy, I doubt if anyone will deny that cars are primarily for transportation (plus conspicuous consumption, but let’s leave that aside), and not for home theater replacements. Yet car manufacturers do put fancy sound systems in. They cost only a small fraction of what the car as a whole costs, but people do expect to see them there. Stastny: I think: Contact, not content, is king (Douglas Rushkoff). Odlyzko: A very nice way to put it. My own preference is to say that connectivity is king. Stastny: It is not only connectivity, it is also identifiers. If you consider P2P networks, also the identification is basically peer to peer (see Skype), at least the trusted identification The only centralized part of Skype is the Identification and Authentication. On a global scale trusted 3rd parties and a circle of trust will be needed This is the reason why Verisign is involved in anything dealing with names, addresses, ENUM and numbers (and The COOK Report on Internet May June 2005 RFIDs) Issues of Security Coluccio: Richard Stastny stated: “If you consider P2P networks, also the identification is basically peer to peer (see Skype), at least the trusted identification” As Internet security experts go, I’m but a step above a total neophyte, I’ll admit, so please bear with me on this one for a moment. It’s good to see you inject the trust component in your statement above, since it seems to have been a principle that’s been absent in many of the previous discussions and sidebars, as I look upstream. How likely do you think it is that renegade applications with large followings will even care about compliance to such criteria, if the communities they serve are okay with the idea of carrying on without it? I ask because I’m not so sure that such an eventuality is all that far fetched, and indeed may even serve the purposes of some by obfuscating their identities that would be obvious in an erstwhile controlled manner. Stastny: Not at all - as long as they do not want money from you. ;-) Most people also do not care who is calling in the PSTN phone, and it is easily possible to make anonymous phone calls (e.g. from any coin box) but in most cases it is possible to identify the calling party (but not the called party). The basis of this is that the telcos can be trusted and also that the telcos trust each other. In every phone call a networkprovided identity is transmitted from the originating office to the terminating office, even if the user has ordered to restrict his identity (CLIR). The destination network (and attached emergency services) can trust that this information is correct, and the originating network can trust that the destination network will not deliver the information to a normal user. One problem remains in fixed networks, the identity is linked to the line (or enddevice), but not to the person. In mobile phones (GSM), the trust has been extended with the SIM-card the network can authenticate the SIM-holder in a quite secure way and therefore can trust that the user is really the account holder. This allows for making payments by using the mobile phone. The called user may also trust the CLI displayed somehow, but not perfectly (more than 50% of mobile users use anonymous prepaid cards - so the network knows which user to trust, but not the end-user) A similar situation is necessary on IP. A way to authenticate and identify yourself to others is needed. Ideally the SIM-card approach is extended. (side remark: SIM cards (to be precise IMSIs) may also be used -since last year - by any provider, not only by mobile operators. ETSI is standardizing for the past year the so-called Universal Communications Idenifier (UCI). The UCI consists of a numeric part to provide a unique identity (e. g. an E.164 number), a alphanumeric part (e. g. your name or role) and additional information - a certificate telling you something about the trust. You may use anonymous UCIs (e. g. aliases) but the recipient knows that this is an alias. The next idea is that the UCI are exchanged between called and calling party, so you also know who is answering the call, which may in some cases interesting if you are calling your bank. And basically UCI is not only for voice calls, where you may identify known person via their voice, but also for all others types of communications. The basic problem here is that you need a global trust network to provide you with UCIs certificates, etc. -which simply does not exist - and some say will never exist. Commercial VoIP Pricing Power Coluccio: Interesting Washington Post article. http://www.washingtonpost.com/ 79 wp-dyn/articles/A31113-2005Jan23.html ?referrer=emailarticle NexTones’ and Neustar’s roles are mentioned, among a cast of others, with the main theme being the daunting challenges that exist for the smaller startup service providers who are first getting into the business of VoIP, both of the enterprise and residential all you-can-eat types. Scrambling For a Slice Of Internet Phone Pie By Yuki Noguchi Investors showed little interest when Richard M. Tworek started setting up Internet-based phone systems for businesses two years ago. Now the company that Tworek founded, Qovia Inc., has $16.1 million in venture capital and 120 businesses use its software to make sure their Internet connections stay robust enough for calls to go through glitch-free. But last week the Frederick company laid off 16 of its 59 employees and retooled its business plan. Like Qovia, most Washington area companies in the Internet phone business are relatively small and still searching for a secure and profitable niche in a fast-changing sector. They range from start-ups that provide behind-the-scenes support for the technology to alternative phone providers, such as Primus Telecommunications Group, that sell Internet calling plans directly to consumers. [Snip] The number of consumers using VoIP is expected to grow to about 4 million by year-end from about 800,000 today, according to Frost & Sullivan. The growth may be driven by major communications companies such as cable giant Comcast Corp., which this month announced plans to offer Internet calling to as many as 20 million homes this year and another 20 million by the middle of 2006. The percentage of business phone lines that make connections over the Internet is expected to grow to 13.1 percent by COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA the end of 2006 from 7.2 percent today, the research firm said. Major corporations such as Ford Motor Co. and Boeing Co. already have traded at least part of their traditional phone system for Internet-based networks. NexTone Communications Inc., also based in Gaithersburg, makes software that helps carriers hand off calls to each other and monitor the traffic on their networks. The company, which has received $32.5 million in venture funding, is trying to win the business of large VoIP carriers but it’s competing against large equipment vendors. The Internet phone trend also has affected companies not directly involved in the field. The primary business of District-based NeuStar Inc. is maintaining the North American database of phone numbers, a service that is essential for carriers to route calls properly. NeuStar has retooled its technology to track the growing volume of calls riding over the Internet, said Jeff Ganek, chief executive of the company. The most visible players in VoIP are the companies that provide and market the new technology directly to consumers. [Snip] The most daring companies may be startups created solely to offer Internet calling to consumers, such as SunRocket Inc., a Vienna company recently launched by former MCI executives. SunRocket markets itself as a consumer-friendly alternative with “no gotchas” such as activation charges and cancellation fees charged by big phone companies. But analysts said such small players are about to face the formidable brand names and marketing budgets of companies such as Comcast. COOK Report: The report “Impact of Skype on Telecom Service Providers” is interesting. http://www.evalueserve. com/download_mediacenter.asp?dfname =Skype&user=N&download=1 Odlyzko: It’s hard for me to get excited by this report. Why do these folks think that Skype will win, as opposed to some other VoIP solution? The range of projections for year-end 2008, between 143 M and 246 M subscribers, is ludicrously small, given all the uncertainties. Also, telco revenues do not have to decline because of VoIP alone. (Migration of voice to cellular, and competition from cable are much more serious threats.) If a telco has a monopoly, and under the traditional European model, say, charged $20/month for basic service including some local calling, $20/month for long distance voice, and $20/month for broadband, all it has to do (subject to regulators permitting, and protests of consumers, ....) is to charge $60/month for broadband with voice tossed in for free. (Actually, it can probably charge $50/month and come out ahead, as it can then eliminate metering and billing costs.) (These numbers are not meant to represent any particular country, just to illustrate the principle.) VoIP, whether Skype or Vonage, or anything else, is a threat to the traditional price discrimination, taxation and subsidy model, but not necessarily to telco profitability. There is a slight threat to profitability, because of the lower ability to price discriminate. (Roaming charges are high not because costs are high, but because people traveling internationally are thought to be more affluent and more willing to pay for their voice calls home, etc.). Now multi-modal competition from traditional cellular, cable, and broadband wireless are a different story. Coluccio: Andrew, after reading all that you’ve written just now, and I myself taking a moment to apply the principles of price discrimination, pricing elasticity and the bundling strategies that are now all too common in the industry, it dawns on me once again that even if the incumbents write down their embedded boxes that continue to produce cupric oxide, they’re still left to face the bondsmen who, for whatever reasons of the past and present, have enabled the telcos, every time an issue comes to maturity, to continue rolling their debt forward onto 80 new twenty year terms. From my point of view, these last minute adaptations to market pressures that the incumbents are now exhibiting, both the ILECs and the IXCs, alike, will not be enough to remove those longerstanding financial burdens that they’ve accrued over the decades, from their backs. I could be wrong about this, but I don’t see how. Any thoughts on the matter would be appreciated. Odlyzko: You may very well be right. Those bonds, deferred tax liabilities, and pension and medical care to retirees obligations might sink them even if nothing else does. Shockey: Think unfunded pension liabilities. They love to bury that stat in their 10-K Plus let’s remember the SBC and BS have paid 40 some Billion dollars for the worst wireless network in the US. Gill: The latest net adds for Cingular are 1.8m. Whatever else they might have done, their network integration is being done extremely well. Wireless Business Models Shockey: Someday Vodaphone will stick Verizon with a 30 Billion dollar bill to buy them out of their share of Verizon Wireless. Gill: Verizon would probably like for nothing better. Verizon Wireless is about 40 percent of revenue and the is by FAR the fastest growing portion. Shockey: So what is their cost of capital? The wire line business is being canabliized by the wireless business and here comes both Wi-FI, WiMAX and Cable which in my opinion gets 30 percent of the residential business by 2008-9 and SIP trunking eviscerates the enterprise market by even greater percentages if the IETF can ever get its act together to define a rational security model for The COOK Report on Internet May June 2005 SIP, etc. Patience. It took how many years for Southwest Airlines to finally crack the business model of United, AA and Delta? Gill: I don’t have the same confidence in WiMAX that most people seem to have. I’m still not seeing how this is actually going to work in real life deployment, especially in unlicensed spectrum, unless _all_ players in the spectrum agree to use the same phy. ISP/Hosting xDSL 5 Cable 5 Customer Acquisition xDSL 15.5 Cable 14.0 CPE xDSL 4.5 Cable 3 Home Installation xDSL 5 Cable 6 Service/Billing xDSL 11 Cable 7 Maintenance xDSL 4 Cable 1 Forster: I’m a little skeptical about WiMAX as well. I assume the technology is pretty good, but anywhere that has two decent broadband providers will be a tough area for a third entrant. It might do OK in un-served areas, although there it will have to compete with lower priced Wi-Fi gear with meshing capabilities. Sege: WiMAX will probably be good for rural areas and small business where cable/fiber does not reach today. Note the success that Towerstream seems to be having in the Northeast with preWiMAX into businesses today. In the long run fiber and wireless combinations are a winning combination, but in the meantime DSL & HFC Cable/ DOCSIS should pretty well saturate the broadband market in much of the US and other similar markets. It will also be good for transport from a city PoP to a metro-scale Wi-Fi mesh. Most of our installations today use “preWiMAX” equipment from various vendors. None are interoperable and all have their quirks. Standardizing this will be useful for compatibility and price reduction. St. Arnaud: I agree with Jim that I don’t see a business case for fixed point-topoint wireless in competing with DSL or cable. A good web site http://www.cybertelecom.org/data/broadband.htm shows the costs of DSL and cable deployment. It is important to note that all these same costs would apply to a wireless provider, but I suspect that CPE equipment and maintenance would be higher for fixed wireless. The only difference is that transport/network costs are already largely paid for in the case of DSL/coax. More importantly, especially with cable, it is very inexpensive to offer higher speeds. In Canada standard cable offering is now 5 Mpbs is $C 44.95 per month, broadband “classic” at 1.5 Mbps is $24.95 and broadband light is $12-$19. Cable and DSL Internet Access Cost Structure - an example $ Per customer per month for model network build: Total Costs xDSL $47 Cable $40 Your comment about the variance in cable speed potentials is well taken, and should remind everyone that the MSOs took the same strategy of holding back on the full potential of the DOCSIS design that availed itself to 30 to 40 Mbps capabilities, initially, just as the FTTP networks of today are doing. By the way, I have a Return on Investment model that was developed here at Tropos and iterated with about a dozen carriers for accuracy. It shows $29 total cost per home passed to build and less than $8per month per sub to operate. This includes CPE devices for in-home access which may or may not be needed depending on construction, orientation of the radios, etc. All based on LA basin home densities. Coluccio: Thanks for pointing out that page on the cybertelecom site, Bill. The unit cost comparisons cited are from a McKinsey and Co. 2001 report, however, and probably require some tweaking at this point, and going forward, especially as the larger MSOs’ fiber backbones come into play, where transport costs are concerned (if I’m reading the intent behind “transport” correctly). Just as the installation costs have probably dropped in some categories due to the plug and play nature of roll your own installations enabled by the availability of DCE/CPE at retail outlets. Transport/network xDSL 2 Cable 4 81 Only Verizon has made any mention, and even then only in a passing way, about its intentions to gradually crank up throughput speeds on their data component to 100 Mbps, over time, and even here their stated 100 Mbps is still conservative with respect to potential. This is because only Verizon, among the RBOCs, is employing a true FTT’H’ design in a number of their city roll outs today. Most of the others, using copper xDSL extensions from field- and curbnodes to their target points, will max out at much lower rates, and will have to share copper spectrum, potentially with a number of video channels, besides. As for WiMAX, this is an area that remains an enigma to me, both from the standpoint of its relative potential from a technological standpoint, and its perceived popularity among SPs going into the future. I routinely become involved in discussions on emerging wireless last mile techs on my forum, and listening to the experts, some of whom are recognized in IEEE circles, has given me a lot ot digest, leaving me still in search of an opinion to formulate on my own. Retzer: I think a lot of the interest is in finding some broadband access method that can inexpensively bypass the LECs and MSOs. Local/regional service providers are concerned that the big boys will muscle them out of the action. Gill: I am convinced that going head to head against DSL/cable is a fool’s game. The wireless value proposition is not in raw speed, it is in connectivity. Play to wireless’s strengths, not telco/cables. Retzer: Thanks, Vijay. I’m not an ISP but work closely with a number of them. What do you suggest for value proposition for local/regional ISPs in areas that are well-served by cable Internet and/or DSL? I think that some, at least believe COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA they need to be able to offer a high-speed with the mag’s editorial director, Stephen ing of new fiber to the home builds are access method. It’s a little scary if you’re Hardy, who committed to me to “pull concerned. small but depend upon service from these some strings” and have it posted as a Instead, they’ve elected to take the slowreally big companies. Having alterna- feature, soon. tives is nice. er route of building out their own service Coluccio: Richard Shockey, with all that territories, complete with the time conMenard: In a duopoly, things should be said, I agree, and it astounds me to wit- suming prospect of delivering on triple unbundled ness the the level of clinical denial that play offerings, at a time when the cable is being demonstrated by some of the operators are stepping up to their front Gill: Jere, the best value proposition RBOCs, as they throw money the way doors with eviction notices in hand. would be playing to the mobility of wire- they’re doing at FTTP. less, as well as trying to be an alternative Over the longer term, the ILECs will to people who can’t get cable/DSL or are Not that I’m against fiber deployments, eventually be forced to deal with comtired of both, though the advantages in mind you, but I don’t see how they can petitors of all sizes whose only means latency and raw bandwidth are hard to continue to spend like drunken sailors, of accessing subscribers will be through surmount, esp. if you have gamers in the after coming to terms with all of the their own facilities buildouts, whether house, where the service lives and dies above. I’m sometimes inclined to think over wide territories or in localized tenby the ping times to the game servers. that it is merely a last-ditch effort by the ant complex situations, whether they be captains of those ships who are trying to fiber-based or wireless, or some hybrid It is entirely possible that there IS no convince themselves that the Iceberg in thereof. This will leave the telcos in a compelling value proposition in dense their scopes is only an optical illusion situation where they have absolutely no areas well served by cable/DSL. Taking (pun intended). And I tell ya, to listen control over those upstarts, whereas, in churn rates from cellular of about 2% of to Seidenberg of Verizon is almost a a contrasting way, they could have been user base a month, if you get one phone religious like experience to behold, the enjoying a symbiotic or even dominant call from a user for support, chances are guy is so good. I’m convinced that he relationship with those startups, while you’ve blown the margin on that custom- actually believes everything he’s saying, serving them as a wholesaler of their own er for the entire lifetime of the customer. *sometimes.* ;) facilities and services, to boot.” This being wireless, it is just going to be very hard to deal with. It was about a year ago that Qwest’s CEO VoIP Pricing Power publicly announced that his company Retzer: Thanks, Vijay. Good response. was broke. At that time he hinted that Encore I think there is definite room for local/ he was seriously considering partnering regional ISPs to provide network and with some of the CLECs, which, in retro- Retzer: My apologies for inserting my web services that the giants don’t neces- spect, probably would have been a much two cents when you asked Andrew, but sarily provide very well. You provide wiser thing to do, compared to squashing I’d also like your views on a couple related points. I was struck recently by some good insights on the limits and them into the ground. Comcast’s conclusion that they could capabilities on the wireless access side. Earlier today I wrote the following in my charge nearly $50 per month for VoIP I appreciate it. SI forum on the subject of the RBOCs when Vonage is all-you-can-eat for $25 Coluccio: It’s not simply a distinction of blowing a good thing (indeed, stepping and Skype is free to call other Skype the technologies that come into play here, deeper into the fire) by vanquishing their subscribers (I don’t, however think most but the business models and the type of UNE-P and fiber facilities UNE-L- based consumers are generally impressed with ensuing operating restraints and costs to competition instead of embracing them the limited network pitch). Now there is pricing power! They gain that pricing end users that any business model usu- as partners: power, assuming they haven’t miscalally imposes, as well. Where WiMAX is generally perceived to be primarily a “In retrospect, one has to wonder about a culated through a combination of adservice provider’s platform, many Wi-Fi potential outcome that could suggest that vertising/name recognition and I guess mesh and grass-root, ad-hoc approaches the RBOCs and most of their sibling and bundling. Everyone knows that Comcast are not, or they may represent a hybrid of lesser ILECs around the country have will be around and if you don’t like the service you at least know there is somethe two, or even more, approaches. taken the wrong approach to DSL and one local (their ads says so), but who are other forms of competition, all along. these Vonage guys (the consumer thinks By the way, I thought of you and Francois when the hard copy of this month’s Consider, where they could have had a even if the current ads are great). So here issue of Lightwave Magazine arrived. ready stable of resellers and distributors is Comcast able to charge $50 a month It’s accompanied by a supplement that of their facilities in the way of CLECs for a service with a marginal cost of is devoted to Blown Fiber technologies and third party agents, they’ve practically probably a few bucks. I’m not even sure and applications. It doesn’t appear on vanquished all of those, for all intents they have to pay the usual government their web site yet, but I’ve been in touch and purpose, especially where the shar- fees. If Comcast could do that, then why 82 The COOK Report on Internet May June 2005 not the ILECs? Where have I, or perhaps Comcast gone wrong here? Is this good public policy? Since Comcast, and presumably the ILECs have considerable market power they face a downward sloping demand curve where a company subject to perfect competition would be a price taker. Perhaps we have, for the moment a market where the major incumbents can charge $50 and known, credible competitors like Vonage can charge $25. The net result of this will be at this stage far fewer people going to VoIP, although the price is bound to drop as others enter the market and consumers gain confidence. Does a slower transition and higher cost for VoIP cost the public or are we just as well off? At least the industry seems to be moving now on VoIP. It seems that we may have finally passed an era where the incumbents were apparently dragging their feet. It seems to me that the Internet and applications like VoIP that use it may be characterized by decreasing long run average costs - that certainly has been the case since 2000, although I’m not sure how much price drops since then have been bubble-related. Do we have any data to support (or disprove) the notion that this industry has decreasing long-run average costs? If so, then what sort of strategies are appropriate? I’ve maintained for some time that rapidly dropping transit costs mean we should be pushing new bandwidth-hogging applications out the door just as fast as we can, which is certainly contrary to recent history. What would the computer and network industry be like today if the PCs had grown slowly? Imagine, if you will that IBM decided instead of going open with the PC they had kept it in-house where it competed with the mainframe and the result was a slow evolution. What would have been the economic impact? Odlyzko: Apropos the last paragraph, I agree. I have been saying for a while that the number one imperative for service providers is to teach their customers how to increase their traffic. And some are able to do it. I was in Korea a few weeks ago, and their traffic (as well as that in Japan) appears to be growing close to 100% per year, whereas here in the U.S. we appear to be down around 60% a year. As for Comcast pricing (whether it is $50 or $40/month), it does not look sustainable for long, but there is a lot of inertia/pricing power/... in the system. A few years ago I dug up some data on banks, and at that time they had a few hundred billion dollars in checking accounts that bore no interest, even though most of that money could have been moved into money market accounts. Some of these inefficiencies can persist for decades. As one example that I have been citing occasionally, if you go to your neighborhood gas station, you are likely to see prices per gallon along the following lines: regular (87 octane) $1.79 intermediate (or whatever it is called, but 89 octane) 1.89 premium (92 octane) 1.99 (This is a common but not invariable pattern.) Well, if you are in the habit of buying the intermediate grade of gas, then you can save easily. Instead of 10 gallons of that grade, get 4 gallons of premium and 6 gallons of regular, for a net savings of $0.20. Reed: Jere - I’m still puzzled as to why you think Comcast might have pricing power? The IBM analogy is underwhelming. IBM at the time was responding to a very aggressive competitor, called Apple, who already had competitors such as RadioShack with the TRS-80. Retzer: Meaning they didn’t have much to lose? That wouldn’t necessarily have prevented them from holding on to their rights rather than let Intel and crowd take over the “IBM PC.” If they had instead kept it in-house in competition with their own mainframes history could have turned out a lot differently. Potentially billions or even trillions in wealth might not have been generated and we could all be staring at 3270 screens. Apples have always been better but the company’s ability to leverage 3rd party developers 83 and market has never been that great. Even today, most corporate folks think Apples are for artists and academics. Radio Shack just seemed to run out of gas. Could Apple have felled the IBM mainframe the way Intel-Microsoft duo did? We’ll never know but I have my doubts. I don’t think I was saying that IBM had “pricing power” in that market. Actually, the point I was making is that incumbents dragging their feet can slow the deployment of new technology and the development of new markets. In the case of VoIP, the ILECS allegedly drug their feet on to prevent more rapid cannibalization of existing PSTN revenues. Strategies like that have a public cost. Make sense? Reed: Apple was growing rapidly in the corporate market, and at the time (Apple II) was open and modular. Did God give IBM pricing power? No, I think it’s something simpler - I was there in the PC industry when IBM entered, and IBM BOUGHT pricing power. They paid flat, upfront contract fees to get the most popular apps on Apple computers ported to the IBM so they would be available on Day 1. In other words, there was very little about the letters I B M that gave them any pricing power in that market whatsoever. They did have the opportunity to enlarge the market by making it seem less risky to use PCs. And that was their long-term edge. However, every time they tried to close the box (Microchannel was a good example), customers dropped them like a hot potato. So I think Comcast’s $50 offering will fail to generate volume, unless they can link it to other things that customers value from them, that they can’t get from Vonage or those guys who are undercutting Vonage (there’s a $200/year for two lines Vonage clone that I just got an ad for. Same service as Vonage, essentially). “Pricing power” is merely a descriptive term for the results of underlying advantages. It doesn’t seem to me that it exists as a “real thing” separate from COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA such things as choke points or barriers to entry. Retzer: I think we agree. It simply describes a firm with a downward sloping demand curve. I would add name recognition and an established business relationship with millions of customers as substantial advantages for Comcast that give them pricing power. Coluccio: Jere wrote: “My apologies for inserting my two cents when you asked Andrew ...” Not to worry, Jere, since your two cents often receives more than the usual two cents in currency exchange. Where did you get that figure of $50/mo for Comcast? Releases I’ve seen put them at $40 ... http://makeashorterlink.com/ ?S26A3115A ... while offering more features. (Hey, I’m only reporting the story here:) But there’s more to their purported feature advantages, I think. Despite their having sunk millions into circuit-switched voice prior to waking up and fully committing to VoIP, they had already begun providing circuit switched services to 40,000 residents and small businesses. Okay, that’s on their dime, and it shouldn’t impact their new biz plan to recover investments in VoIP, but rest assured it does. Also, there is always the spectre, at least in the back of my mind, that their new entry into a Packet Cable-based variation of VoIP will far outshine at a perceptual level if not a real, measurable one, from the perspectives of reliability and assured quality through the application of QoS, a better product than that of Vonage, or CallVantage, or Lingo, to name just three of a growing number of look-alikes. Lest we forget, it works counter to the Cable Operator’s interests to enhance the quality of best-effort parasitic VoIP offerings, while their own QoS-enabled voice services share the same cable modem space as tenants in the same portion of the spectrum. And taking an even more sinister view of the matter, there is nothing forcing MSOs to even support the parasites, much less enhance their carriage. set its product apart. I don’t know if that equates to the equivalent of what you refer to as pricing power, but it sure suggests a level of market power, even if measured in another, perhaps even a Machiavellian, dimension. Officials of the company - which plans to charge $40 per month, or $5 to $20 more than most of Internet phone services - said Monday they believe customers will pay more for features such as battery backup to keep the phone line running during power outages. So, as long as Vonage and its peers lack any means of supporting a suitable level of quality with consistency across all of its venues of operation, the MSOs will have, from a consumer experience point of view, a superior product, and as such would normally allow them to enjoy a premium, whatever it might be in terms of a percentage or dollars per month. And for these reasons I believe we’ll see the MSOs experience internal tensions between the throttling up of their absolute cable modem data throughput capabilities, on the one hand, which could potentially improving the quality of competing, parasitic voice offerings, and the holding back on those speeds, which would have the detrimental effects of making them appear non-competitive with future FTTP throughput capabilities. Later Coluccio: Correcting my earlier URL posting of the Comcast PR, the press release below discusses the Comcast VoIP offering I referred to, earlier. The URL, followed by the text for posterity: http://www.rgj.com/news/stories/ html/2005/01/10/89432.php?sps=rgj. com “Comcast to offer VoIP service” by Maryclaire Dale ASSOCIATED PRESS 1/10/2005 09:45 pm NEW OFFERING: Comcast Chief Executive Brian Roberts delivers the keynote address during a telecommunications conference Monday in Phoenix. PHILADELPHIA - Comcast Corp. is joining the crowd of major cable TV and telephone companies venturing into Internet-based phone service, but with a higher price and extra features which the nation’s largest cable provider hopes will 84 The new Digital Voice service won’t be immediately available beyond the three markets where Comcast has been testing it. The company plans to offer the service in 20 of its markets by the end of this year, and the rest of its territory during 2006. That delay that could prove costly in a young market already marred by price wars among AT&T Corp., Verizon Communications Inc., Vonage Holdings Co., and a growing list of no-name rivals. But despite the apparent disadvantage, Comcast officials said they hope to sign up about 8 million phone customers, or a 20 percent share of the markets it serves, within five years. [Snip] VoIP and Security in Government and the Enterprise COOK Report on January 27: http:// www.computerworld.com/newsletter/0,4902,99258,00.html?nlid=PM JANUARY 26, 2005 (COMPUTERWORLD) - A new report from the National Institute of Standards and Technology urges federal agencies and other organizations to take care in switching to voice-over-IP technology because of security concerns. The 99-page NIST report, “Security Considerations for Voice over IP Systems,” includes nine recommendations for IT managers to help them implement VoIP in a secure manner. “Lower cost and greater flexibility are among the promises of VoIP for the enterprise, but VoIP should not be installed without careful consideration of the security problems introduced,” the report says. The COOK Report on Internet May June 2005 “Administrators may mistakenly assume that since digitized voice travels in packets, they can simply plug VoIP components into their already-secure networks and remain secure. However, the process is not that simple,” the report says. The report, authored by NIST computer security experts Richard Kuhn and Thomas Walsh, as well as Steffen Fries of Siemens AG, appeared in draft form last June and was formally released in final form earlier this month. Today, NIST included excerpts from it in an email newsletter. Among its recommendations, the report calls for building logically separate voice and data networks where practical, instead of building a single converged network. It also calls for using VoIP firewalls and routinely testing them. Another recommendation says that “if practical,” VoIP softphones should not be used where either security or privacy is a priority. A softphone involves using an ordinary PC with a headset and special software instead of a typical telephone unit. Many analysts and even VoIP hardware vendors have discussed VoIP security for years, but the predominant thinking seems to be that such systems can be installed in a secure way (see story). [snip] Coluccio: I’ve see the NIST Report Computerworld summary, and a Powerpoint presentation of same, but not the full report referenced anywhere, itself, so here it is: “Security Considerations for Voice Over IP Systems Recommendations of the National Institute of Standards and Technology,” D. Richard Kuhn, Thomas J. Walsh, Steffen Fries Special Publication 800-58 January 2005 http://csrc.nist.gov/publications/nistpubs/800-58/SP800-58-final.pdf COOK Report : This paper is also worth looking at. http://www.simson.net/ ref/2005/OSI_Skype6.pdf. It is a study that Simson Garfinklel did for the Soros Open Systems Institute. It sparked the following exchange on Dave Farber’s Interesting Persons Mail list. First Commentor: Is Skype secure? The answer appears to be, “no one knows.” The account accurately reports that because the security mechanisms in Skype are secret, it is impossible to analyze meaningfully its security. Most of the discussion of the potential risks and questions seems quite good to me. But in one or two places the report says things like: “A conversation on Skype is vastly more private than a traditional analog or ISDN telephone,” and ““Skype is more secure than today’s VoIP systems.” I don’t see any basis for statements like this. Unfortunately, I guess these sorts of statements have to be viewed as blind guesswork. Those claims probably should have been omitted from the report, in my opinion -- there is really no evidence either way. Fortunately, these statements are the exception and only appear in one or two places in the report. David Wagner: The basis for these statements is what the other systems don’t do. My Vonage VoIP phone has exactly zero security. It uses the SIP-TLS port, without encryption. It doesn’t encrypt anything. So, its easy to be more secure than that. So, while it may be bad cryptography, it is still better than the alternatives. Unfortunately. Adam Shostak: I don’t buy it. How do you know that Skype is “more secure”, let alone “vastly more private”? Maybe Skype is just as insecure as those other systems. For all we know, maybe Skype is doing the moral equivalent of encrypting with the all-zeros key, or using a repeating xor with a many-time pad, or some such? Without more information, we just don’t know. I’m sorry to pick nits, but I have to stand by my statement. No matter how atrociously bad other systems may be, I don’t see any basis for saying that Skype is any better. It might be better, or it might be just as bad. We don’t know. 85 Dave, I’ve been following the Simson/ Skype thread on IP and I’ve read the Columbia analysis of the Skype protocol (http://www1.cs.columbia.edu/~library/ TR-repository/reports/reports-2004/ cucs-039-04.pdf) I’ve known Simson for 14 or so years and have a ton of respect for his technical skills. However, I think there are some significant Skype vulnerabilities and associated legal ramifications that Simson did not discuss in his article. Security is based on trust of the parties exchanging information that they are who they claim and that the data exchanged appears to be random to an untrusted observer. While Skype’s use of encryption supports the second part of the definition, it does not support the first. Because it does not support the first, it is very easy to use the Skype network to intercept communications between any user or to pose as any user. This presents a problem as against both third parties and governmental agencies. A critical part of the Skype network is the “super-nodes.” According to the Columbia paper, super-nodes perform three functions: * Designating the login authority * Media packet forwarding * Routing user search requests Supernodes appear to “volunteer” to perform the function. Or put another way, they are nodes that are not under the control of Skype, but they perform all the routing functions necessary to discover a user and exchange information with the user. Super nodes run on any machine running the Skype program and the machines under Skype control have no way to determine if the super nodes are running unmodified Skype code. If one were skilled in reverse engineering x86 code and one were willing to violate Skype’s user agreement, one could create a Skype node that volunteered to be a super-node. It would appear to all other Skype nodes as a normal super-node. It would perform all the functions of a Skype super-node. However, it would do a little bit more. Let’s call one of these COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA super-nodes a “Bad Seed.” The Bad Seed could point users to another authentication server. Thus, the user would exchange username and authentication information with a “bad relay proxy” rather than the Skype server. That permits the “bad relay proxy” to deny Skype access to a user that I designate. Okay a denial of service attack is not great stuff, but for businesses that rely of Skype (http://news.com.com/No-cos t+Skype+strikes+chord+with+businesse s/2100-7352_3-5553053.html ), having the prospect of a DoS attack could be an issue. Further, the “bad relay proxy” could collect username/password challenge (I’m assuming that Skype’s not sending the actual password, but performing a challenge/response method of verifying the password) data and do dictionary attacks on the passwords. This isn’t a hardcore vulnerability you say. Yep... I agree. The Bad Seed routes some of the media requests if the two peers cannot see each other directly. Skype claims that because the packets are encrypted, the super-nodes are just routing agents. This is true unless the Bad Seed is part of the key exchange (see the next paragraph.) If the Bad Seed is part of the exchange of private AES keys that are used to encrypt the voice data, then they are able to decrypt the audio or text streams. Yikes. If a Bad Seed was a man in the middle of the private key exchange, then the Bad Seed could record your conversation with another user. Okay, that’s not good. Given that any Skype node can become a super-node just by raising its hand and a skilled hacker can re-engineer a Skype node to perform bad acts, then if you connect to the Skype network, you don’t know which nodes are listening in on your conversation. But wait, you say, the requirement for the above bit of scariness is doing a man-in-the-middle attack on the encryption key exchange. You’re right. And here’s where the Skype network is totally insecure. One function of the super-node in the Skype network is to route and respond to user search requests. If I want to connect to “other_dude” on the Skype network, my client sends out search requests to a series of super-nodes. The super-nodes either respond with the address of “other_dude” or forward the requests to other super-nodes. If one of the super-nodes is a “Bad Seed”, that node can respond that it is “other_dude.” Because there is no cryptographic trust or any form of trust authority in the Skype network, any super node that returns the information about “other_dude” is trusted by my node. so the police don’t hassle you. But you can do it. An aside... SSL certificates are signed by a trusted third party. That third party validates that the certificate is held by the organization that claims to hold the certificate. Using SSL insures that the party that I’m communicating with is the party that they claim to be within the bounds that I trust the signer of the SSL certificate *and* that once the connection is established that no one can understand the data exchanged with this party. That initial trust of the signed certification is a critical part of the security of the overall communication. If I do a session key exchange with an unknown party, the communication is *not* secure. This is the case with Skype. SIP is different. SIP supports encryption, but most SIP providers do not make use of it. The Microsoft SIP client libraries have the option of communicating with the SIP server via TLS (TLS is like SSL, but uses the same IP port for both encrypted and unencrypted traffic.) Additionally, the media portion of a SIP call can be encrypted by setting a flag in the media descriptor. While most SIP providers do not use this functionality, it’s part of the SIP spec and can be turned on. Note that the machines that could play man-in-the-middle with an encrypted SIP call are controlled by your SIP provider (rather than any machine running Skype.) Thus, you can trust the security of your call as much as you trust your SIP provider. The Skype network relies on trusting the super-node. A Bad Seed can perform a man in the middle attack during the session key exchange by posing as the party being contacted (or forwarding the information of another compromised node) to a caller. So, my Bad Seed is able to route call requests to an untrusted node and do a man-in-the-middle during the key exchange and snoop into my call. The only question is how many Bad Seeds to you need in order to capture a significant percentage of the routing requests that go over the Skype network. My guess is that the number is in the hundreds. So, with a hundred machines located around the world, I could intercept any Skype call and record it. Pretty scary. The PSTN primarily uses pairs of copper wires to transmit voice communications from my house to the phone company central office. I can gain physical access to those copper pairs very easily as long as I have physical proximity to the location of the person I want to snoop on. It’s not hard to do. Yeah, you have to paint a white van with “Verizon” or “SBC” 86 If the government wants to do it, it’s somewhat harder. The government has to get a warrant to listen to your phone conversations. Once they obtain a warrant, they present it to the phone company which makes an entry into their switch to record the call or send a real-time copy of it to the government. With unencrypted SIP calls, if you are able to intercept packets, then you can tap the call. Anybody on your LAN can listen into your call. This level of security is no different than anyone in your house can listen in on your phone calls and anyone in your office can probably do the same. Anyone who can intercept the packet stream outside your LAN can also listen in on the conversation. This is more of a challenge. UDP packets (the stuff that the media stream goes over) may or may not be routed through the same backbone during all parts of the conversation. There is a certain amount of security with the packets going over the backbone. The ability to snoop on an unencrypted SIP call is marginally more difficult that snooping on a PSTN call. For the government, it’s more of a challenge. Because the media portion of a SIP call goes directly between the end points without going through the SIP providers network. This raises an interesting issue: http://news.com.com/2100-7352- The COOK Report on Internet May June 2005 5296417.html. This is interesting for two reasons. First, SIP “telephone” companies like Vonage will have to provide a flag to allow them to intercept the media stream and decode it if the government has a warrant. Second, the government has acknowledged that SIP callers have the same expectation of privacy that copper-pair PSTN callers have. This is really important. Users of peer-to-peer file sharing programs don’t have an expectation of privacy in their use of P2P programs. That’s why so many folks are being sued (http://news. com.com/RIAA+files+754+new+fileswapping+suits/2110-1027_3-5494259. html?tag=nl ) Skype touts themselves as a P2P voice communications system (http://Skype.com/products/explained. html). That means that if you use Skype, you have the same expectation of privacy as a P2P user. Given that the government has the resources to build Bad Seeds and that P2P users have no expectation of privacy, you can bet that there are government run Skype nodes looking for Skype communications between Osama911 and Sleeper_in_Seattle and that the government doesn’t have a warrant for these activities. To conclude my long rant, the Skype network is radically insecure because it relies on untrusted super-nodes to perform trusted functions, most notably user look-up. It’s easy to build a compromised super-node (a Bad Seed.) With a limited number of Bad Seeds, the communications between any users can be intercepted or denied. It’s something that a person with the resources to rent 100 servers in collocation facilities around the world could do (that’s about $10,000 per month investment.) Given that Skype is a P2P network and users of such networks are not afforded the same expectation of privacy that users of the PSTN and other telephone networks are afforded, the government could use such a mechanism to listen to Skype-based calls and have a reasonable legal argument that they do not need a warrant to do so. That’s my 2 cents. Thanks, David P. S. -- I was CTO and VP Engineering for an Internet security company for a number of years and I’m a member of the Rhode Island bar. Garfinkel: Hi, David. Lovely to hear from you again. I actually specifically discussed this attack with the folks at Skype when I wrote my piece. They say that all authentication and authorization communications are encrypted with a public key, and that the private key is only at the Skype HQ. I’m not sure where to take that idea, but it’s an idea. My piece doesn’t go into details of this kind of attack because it was written for OSI grant recipients who are not technical. These grant recipients want to know if they can use Skype or if 87 they should just keep using their analog phone lines. Their primary concern is being wiretapped by the government in the host countries. These grant recipients and other Open Society activists have heard so many crypto-nuts saying “Skype isn’t safe” that they’re using the local PTT phones instead. They’re pouring their grant money into the coffers of the local governments (who own the PTTs) and they are opening themselves up to systematic eavesdropping. This is just a completely different threat model than most of the anti-Skype people are even aware of. I’m sure that if David wants to write a standards-based SIP system and distribute it to the OSI grant recipients, and make it as easy to use as Skype, and make it sound as good, and make configuration instant, and make it interconnect to PSTN networks for a few pennies a minute --- I’m sure that if David wants to do all those things, then the OSI grant recipients and other Open Society activists will be happy to use it. But those activists live in the real world and have to make real choices. Right now the choice is using Skype or using the analog phone on their desk. Discussions about theoretical vulnerabilities and badseed super-nodes just scare the activists into thinking that this Internet security stuff is too complicated, and they’re better off just using that analog phone. Symposium Discussion February 6 - February 24 Skype, SIP, the Enterprise and Security Skype is Like Apple II in the Enterprise Highlights far better than boundary firewalls, and sandboxing (virtual machines and chroot jails) is far more relevant than “shameful packet inspection” in gateways. Reed: Skype is kind of like the Apple ][ in the enterprise. People are using it, but the IT Department doesn’t like it one bit. Editor's Note [on March 25]: Wrestling with a final draft I have just chopped 13,000 words of rather detailed technical discussion. I had put all of this through an extensive edit and format. However 120,000 words is simply too much! (And the 120,000 words was edited down extensively from the list discussion!) I realize now that it is an example of reaching the limits of discussion with 40 contributors over 3 and one half months. I’m sure the Skype people want to see it viewed positively in the enterprise. I’m sure the Skype competitors want to spread unreasonable FUD, and also justifiable skepticism, in order to keep Skype out of the customers they think they rightfully “own”. The best way of course would be to make their service work as well as Skype out of the box, and not tie it to enterprise sales [As I left Lotus in 1992 to join Interval Research, one of my parting warnings was that Lotus Notes *must* be purchasable and operable as a solution that wasn’t just an “intra-company” solution, or they would lose big to the Internet. I don’t think they listened carefully - they continued to create private networks for customers, and created boundaries for inter-corporate Notes connectivity that were operationally too hard to surmount. This failure was one of the sources of the “Reed’s Law” idea, since it stunted Notes’ potential precisely due to the lack of paying attention to how group-forming creates value]. The reasoning behind “boundary security” is seriously flawed - connectivity across the corporate membrane is the MOST business-critical connectivity there is, not the least - and at the same time, putting your protection at the boundary ignores the nature of the worst threats, which are predominantly INSIDE the firewall. The best protection is right up next to the application, and based on authentication at the application semantic level. That means that per-machine firewalls are Retzer [reacting to what I have just cut]: Another [outcome], perhaps is to deploy a stupid network and not worry so much about what goes through it. Coluccio: Maybe we’re sitting too far apart in the classroom, Jere, but I can’t tell whether you meant that seriously, or not. In any event, you’ve hit on a growing part of the overall problem by highlighting the dissimilarities between that which [in networks] is rote stupid, and that which aspires to act in a stupid way. It only becomes more so when you consider the options now taking advantage of stupidity to new heights, a la the Skype application whose presentation we just listened to. The situation approaches that of a Gordion knot when you consider “what is” and where one would like to be, with the backdrop being one of the only forms of Internet based - or PSTN-based, for that matter - services that is still being monetized beyond break even levels, in any appreciable way. And of course I’m speaking about voice and video conferencing, both of the measured minutes variety and the all you can eat, as well. And so, the translation devices at the edge make everything appear homogeneous. Today we see another example of this by way of the NAT-firewall traversal 88 capabilities announced by Tandberg, as illustrated in this NW Fusion article that arrived in my snailer today: http://www.nwfusion.com/news/2005/ 0207tandberg.html As a non-ISP, but one who is well accustomed to the kludges that result from firms merging and being acquired, my observations suggest that most of the “dumbing down” through such measures at the edges of the network that is being accomplished by these devices are an imperative, if not merely lending a form of license, to those who’d prefer - or are forced into - proliferating the existence of heterogeneous environments. And maybe this is sometimes due to service providers being forced into situations, like the example I just used above concerning mergers and acquisitions, or because of a philosophical preference an SP has for selecting “best of breed” software, appliances and network elements for each point solution. Thereby perpetuating the need for such solutions even further down the road, while effectively providing a form of regenerative feedback to the syndrome. "What to do?" is probably not a question that any individual can answer for the whole, but rather, "what will folks do?", is more apt a question to ask here, where matters or security, if not overall architecture, are concerned. And through all of the complexity that is usually ascribed to the newly engineered, deliberate forms of kludges that result when additional boxes are inserted in the diagram, in order to assuage the problems brought on by the other additional boxes, may there not be an added level of security from viridae and worms sent by the bad guys that now have to negotiate those additional boxes, than a purer form of end to end model would present? Although this sounds silly, the question is legitimate, in my opinion. The COOK Report on Internet May June 2005 I’m just asking some questions here that happen to be surfacing in my mind as I read through the list comments, without any prior suppositions of what the answers to those questions might be. From the standpoint of doing any type of business in a commercial context, voice and video conferencing applications may be viewed as being as far removed from the architectural ideals of the Internet as a computer numeric control application is on a factory floor. It’s merely a business tool that has to work right when called upon every time. And they are nothing more or less than that, from the standpoint of most who use them. and Late Adopters are the huge bulge in the frequency chart that is definitely negatively skewed. Davis: Well said, Frank, at least from my perspective. I am an engineer and an architect. I am also the one my company trots out to government and enterprise customers to give the hows and wherefores . . . customers who are highly concerned about cost, availability, insist on the telco-like SLAs to which they are accustomed (four ninies). Frequently there are competing customer groups: the older, telco-accustomed, frequently proxies for their customer favorites and personal relationships built up over decades, and on the other side, younger “whiz-kid wana-bees” who tout their protocol religion and vendor of choice. To some objectors, citing that IP was designed to be connectionless, I respond that TCP was designed to be “connection-oriented”, just as ATM is connectionless but “connection-oriented.” To those who insist, with what I always hear as protocol religion, on EoIP (nothing less than Everything over IP), I respond, and Yakov responded, that IP was never designed to do anything with voice or video, whereas ATM was, and . . . These are real issues up at the political end and tribal end. I am the one usually brought in at stalemate stage. These larger organizations, unlike individuals who are free to care or not, who are constrained by statute, case law and GC interpretation as to liability and contracts, and policy to provide at least disciplined due diligence on security. At the Operations level, the security issues are not only of privacy and confidentiality of digital information asset stores, but the requirement to keep the network up and operational, especially as the post-BUST cost cutting has driven staff to what I think is excessive thin-ness. The interoperability issues that Frank raises are essential in designing and operating. As a simple case, the Geoffrey Moore curves on market adoption of new technology hold true. The economic case is rarely won by the swift. The Middle To meet the “four nines” SLA, absent any, using Gordon’s term, “disruptive” and rapidly adopted by a super-majority of the economically significant population, signaling or other in-band or out-ofband IP protocol, such as Yakov Rektor and some of us techno-agnostics at Cisco attempted to promulgate, the SS&/AIN network is essential. Government-toBusiness, Government-to-Citizen, Business-to-Small Business, and Business-toCitizen will require it. if the engineering and architecture of communications networks is about “the communication” part - User Adaptation Layer, let’s work on that problem, using whatever tools we have or can design in the process. Melissa - who worships no protocol godlet. Is the Enterprise is in Gridlock? Reed: Don’t bet on the Enterprise markets for any of this, Gordon. Enterprise use of new technology is now dead. The Chief Security Officer outranks the CIO, and each one says: “Be afraid of novelty, be very afraid.” Just imagine your worst nightmares and project them on new technology because it’s weird. This just reminds me of 1986 at Lotus when the Chief Security guy (our corporate counsel) said that we should not allow our employees to interconnect our electronic mail with the Internet, because 89 that would be the death of Lotus due to the raging security risk. As a vice president in charge of developing technologies and products around (among other things) electronic mail products and stuff like Lotus Notes, I and others argued that we couldn’t afford not to interconnect. This was not an easy fight. There was “no business benefit” to interconnecting to the Internet in those days - we were told we had to limit interconnect to a small number of essential R&D personnel who could demonstrate a “need to email”. When we engineers said: “No, Jim Manzi should be using Internet mail.” People thought we were joking. Jim, fortunately, understood the point, and chose to make an example of himself. It also reminds me of the time Starwave and some Interval Research people demonstrated the first commercially successful Web service to Microsoft’s executives (ESPN Sportszone) in late 1993. Microsoft could NOT connect to the Internet, again because of corporate policy. We had to do the demo at a site outside the corporate campus because of their overwrought policy. The CSO’s are now feeling more than ever empowered (perhaps by the fear campaign launched by the right to keep America in line with the Bush administration) to stop any technology that might make corporations more productive or better connected to their customers. They are abetted by technology suppliers selling gewgaws pandering to their fears (check out some of the new gear pandering to the fear of 802.11, cameraphones and USB memory sticks being sold today, which costs a lot and fixes nothing). So if I were to bet on the source of the next technology revolution, it would NOT come from the Enterprise Market. It should, but it will not. Not until the CSO’s understand that the best way to secure a company is to stop dealing with customers or suppliers, and build up that moat until they can’t feed their employees, much less help send their kids to college. P. S.: Lest you think I am against secu- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA rity, let me point out that the best petri ... are firing offenses. Retzer: Frank wrote: dish for viruses is the corporate intranet, because everything operates wide open And to back it up, various gear providby default, and hidden from public view. ers are providing traps for such “evil I recently found myself in one such situaI am frustrated that those who argue from technologies” that can be deployed on tion in my own den, at home, where I was Skyping out over my cable operator’s security actually create the most vulnera- corporate campuses. ble and fragile systems. Security matters. cable modem connection to a centrallized CSO’s (non-technical lawyers mostly) So yeah, those of us who live largely enterprise conference bridge, into which are suckers for snake oil salesmen and outside the “enterprise” as consultants, my customer’s Cisco VoIP gear also atnever have to justify their actions in pundits, small-business owners, etc. tached, and into which other participants from home were also being SIP’ed in, terms of business risk/return tradeoffs. don’t see this phenomenon. simultaneously. Coluccio: David, while many of your But I can assure you, you may be able to points are well taken, I don’t know how technically route Skype out of your PC How do you bridge Skype, SIP and you or anyone else can any longer dif- on the corporate network today, but the PSTN? Thanks ferentiate today between enterprise, con- trade organizations of corporate legal sumer, and the caller who is Wi-Fi’ing departments are telling all their mem- Forster: Voice bridging can exist in any his way via VoIP into a conference center bers to make policies against it, to re- signaling/transport domain (PSTN, SIP, via VoIP from a cafe hot spot, through quire technology to block it, and asking Skype); mixed mode bridging most likely the use of a dual-mode cellular device. I senior executives to direct against it. happens in the PSTN (G.711 on DS-0 recently found myself in one such situawith DTMF & Robbed-bit signaling) tion in my own den, at home, where I was After all, you wouldn’t want your em- domain because pretty much everything Skyping out over my cable operator’s ployees actually to be empowered would can connect to the PSTN. The dial bridge cable modem connection to a centrallized you? That’s a scary thought! itself looks like a bunch of phones and enterprise conference bridge, into which simply does the right mixing. my customer’s Cisco VoIP gear also at- This fear of novelty is one of the many tached, and into which other participants reasons why genuinely disruptive new Coluccio: Dead on. And to answer Jere: from home were also being SIP’ed in, communications technologies come last simultaneously. to the Fortune 500. The enterprise mar- As Jim has already noted, the least comket eventually gets there, but it’s rarely mon denominator is a plain old PSTN The mix of voice applications and point in the vanguard of anything. compatible 800 number conference bridge solutions currently being used and on the - although I’ve been told that if the client drawing boards has evolved to a level Coluccio: David - Just for clarity, I’m so elected, they could also have used a that is far too incestuous to allow such Skyping from home, not my clients’ of- VoIP feature accommodation in their IP distinctions - or exclusions - to be made fices. ;-) PBX to achieve the same end. Everyone when discussing matters of security, soledials into the bridge which effectively bely on the basis of who, or from whence, On that note, however, I recently spoke comes a DS-0 exchange, so at some each call sessions are being set up. with one my associates who IS situated session is being set up through a gateway behind a client desk (and who also VPNs of one form or another. I used my SkypeThese dynamics will further frustrate from home two days out of the week), out feature from home to dial in to the future efforts to distinguish between call- and I asked him if the client’s security bridge. The IT staff at the client site used ing classes, as the diffusion in the work hawks were still adamant about prohib- SIP-based phones, which, through PRI place, accompanied by decentralization iting IM’ing over the corporate LAN, gateways to the PSTN, did the same. of larger corporate offices, continues. So, as they had been a while back. I was And the work-at-homes equipped with it doesn’t surprise me why a vertically in- advised that they are now using an in- SIP phones use the enterprise VPN. Altegrated proprietary solution, something house IM package, which, while capable though, I’m not sure of the mechanics like Skype, say, might be so appealing of talking to other vendors/SPs’ versions that are used, but I can find out and report to some. of IM, still cannot do so for the types back to you if you wish. of reasons you’ve cited and otherwise Reed: Frank - I wish that the internals of implied. Also, Skype is not to be seen I think that this should also serve to point enterprise networks were just part of the anywhere on the horizon in that particu- out that, under the conditions I’ve outInternet. lar account, while there are staff that I lined above, at least, different species of have situated elsewhere who say that VoIP don’t necessarily have to touch each However, as I mentioned, in many com- it’s being used with mixed resistance, other in an organic way, but in some way panies, the CSO has ordered that use of still, and I’d imagine that the smaller the must be made to talk a LCD language, contact managers (like Plaxo), VoIP (like number that exists following the name nonetheless. At the same time, the DS-0 Skype or Vonage), access points, Wi-Fi, FORTUNE 500, the more resistance one normalization also serves to dispel any heightened concerns surrounding each of will find. 90 The COOK Report on Internet May June 2005 the variants also infecting one another, since it also serves as a means of invoking an active state under quarantine, if you will. On February 4 Davis to David Reed: Your statement immediately below, to wit: "Don’t bet on the Enterprise markets for any of this, Gordon. Enterprise use of new technology is now dead." "The Chief Security Officer outranks the CIO, and each one says “be afraid of novelty, be very afraid”. Just imagine your worst nightmares and project them on new technology because it’s weird." This is different from the one in your second email below: “However, as I mentioned, in many companies, the CSO has ordered that use of contact managers (like Plaxo), VoIP (like Skype or Vonage), access points, Wi-Fi, ... are firing offenses." While the former is just not true, not in my experience with my customers, and invocation of the universal premise outside of formal logic is usually hyperbole, the latter is undoubtedly true at least for some, and the use of “many” depends on one’s scalar of what many means. I know of none, which only means that this writer knows of none, attacking Plaxo. All of my big customers are doing test beds or actual deployments of Cisco’s VoIP or Nortel’s VoIP. Vonage is, among my customers, in no way singled out as a target. I wish you would explain the point of why you wish, as you state below, that “the internals of enterprise networks were just part of the Internet.” Why should they be? The internals of the enterprise networks are privately owned and operated, under the constraints of the marketplace of P&L, for the business requirements of that business/enterprise. Those enterprises are governed by statutes (Sarbanes-Oxley, HIPAA, etc. Government networks must meet the requirements of Clinger-Cohen, e-Commerce, e-Auth, and coming is HSPD-12.) All businesses are bound to protect privacy and integrity of both financial transactions and employee personal information (personnel, discipline, evaluations, labor grievances of some kinds, EEOC violations). Businesses and Government networks institute policies for very secure authentication and communications with and about employees with some risk, likelihood of being targeted by hostile agents (kidnapping, etc). Businesses and Government networks must maintain privacy, integrity, and transaction logging of sensitive management, trade secrets, intellectual property, strategy information. ered workers” without themselves becoming disemboweled in the process. I would like to hear your dissertation on the compelling reasons for opening all that up to any old curious person, or expose that non-public but legally obtainable information for public discussion. The NIST report - Yes,Gordon, the 100 page one- lays out the risks. No one says the risks are proscriptive. One can, and CSO’s do have to accept risk. I will tell you that I am, with others, engaged on ferreting out an application you did not mention, specifically “GoToMyPC”: a remote control program that defeats and renders useless enterprise firewalls, IDS/IPS, and other packet inspection devices. GoToMyPC, once installed on an internal enterprise machine, maintains an open outbound connection HTTP (443) to a remote proprietary GoToMyPC server and exchanges shared private keys with AES128. The remote agent contacts that broker/server from some outside point with unknown and unknowable authentication standards (weak passwords, e.g.) This places anyone with the legitimate credentials (possessed by a user for use on the internal network or possessed by a poseur who has stolen the credentials via RAT/keystroke logger, or zombie via RAT) in complete control of the internal machine, its access to other machines and network stores. Internal security policies have no effect on restricting the export of internal information as, to the network devices, the request is received from a device internal and local (the invisible remote user not being visible). The encrypted connection between the internal machine and the remote machine through the GoToMyPC broker is opaque to any known sensor or firewall device. Every customer I have wants “empow91 VoIP presents special problems in the corporate/enterprise/government environment. Business cases have to be made, risk assessments done, mitigations proposed-tested-deployed that meet cost-benefit decisions. Information may be free, or practically free on the Internet, but information is not free in the corporate/enterprise/government environments. There, information costs to produce, and that cost has to be justified with some judgment of ROI. The issue for techno types like us is to make the business case. Not by technology alone will one sell anything much any more. Stuart Henshall and Skype Voice Mail On February 9 COOK Report: Please welcome Stuart Henshall to the list By way of background Stuart told me that for a number of years in the 90s he worked for Stewart brands global business forum helping various companies do marketing turnarounds. If I understand him correctly he has been more of a marketing consumer goods person than a telecom specialist. But he is into blogging and he said the blogging world made him away of Skype instantly. He downloaded Skype on day three of its release and his blog appears to be almost all Skype all the time. Reading from his blog and talking with him and using Skype more and more myself makes me appreciate what David Reed wrote here in early December about how Skype is CHANGING fundamentally the way people communicate in general and by voice in particular. At http://www.henshall.com/blog/archives/001095.html we find COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Skype Journal Voice Messaging Fast to share what I was doing with those For that expenditure, I have 4 wireless Forward January 23, 2005 11:21 AM that found the line engaged. I got a lot phones dotted around the house, can call of voice mails. Mostly blank and short for “free” (at least on an incremental I was granted 10 opportunities to share voice mails. I also got VM’s from people basis), and can call anyone without havSkype VM with others. I shared it with that clearly didn’t know how to stop ing my PC or a separate PC server up Buzz and I have this wonderful mes- them... I think that is the “Oh no!” what and running. sage back from him asking for 100 more is it doing factor?.. playing message and invites. I know he actually wants 1000! then asking to record. Too scared to do I then keep coming back to Skype to see Of course, I wanted to share this news anything they end up as 1 minute of si- what I am missing - yes, I could set up a and send it back to Skype . It’s the sort lence. Having added VM to this account I PC that runs as the host machine, I could of voice message people get a kick from. found that there was no way to turn it off. get (for $150 or so) a phone with a USB, In fact in my life I find I don’t forward Why would one want to turn it off? Well and could probably manage to wire it many messages. In the cell phone world in this case a “busy” signal would have around the house to other locations etc., I have a feeling that it costs me money provided better feedback. Now that the I could get Out to call people in the US and then I just don’t have the feature set client is completely offline it has prob- or UK, could work through the voicemail on my landlines. This seemed the perfect ably taken VM from frustrated former system that is in beta, could perhaps get message to forward. listeners all day. Most will be a second or (at a cost) a IN number so that people nothing at all. Thus this account would could call me (paying by the minute?), So, I tried to record it to a .wav file a be better served by remaining perma- probably cant make emergency calls couple of different ways and was unsuc- nently offline rather than collecting VM’s from it, and now my telephone service cessful. As I’ve already sent a voice mail until this trial ends. Collecting VM on an is dependent not just on the power and while in a conversation with another account that someone may not use again cable modem, but on my PC not freezing person. I thought I’d see if I could play seems like a bad idea. as well. Would my cost be less than $20 it back in a conference call. That failed. per month - I doubt it. I also tried Windows Sound Recorder [Snip] with Virtual Cables etc. Didn’t work - at I’ve then thought about using Skype least not yet. In the end my solution was Steve Heap: I sometimes think I am from a PocketPC to avoid using a mobile to send it to my iPod with iTalk recorder really missing the excitement here in - yes, it works, but if you have already out the headphone jack and then iTunes these recent emails about Skype, and decided to have a mobile for other concopy it back to my PC. Clipped in Sound I cant fully put my finger on what is venience reasons, it is probably only for Recorder and convert to mp3. A little too missing. I don’t think I am viewing the international calls that it would be useful, hard. Still I have an mp-3 and now can world through my many years of telecom and there are ways to do that - such as provide the feedback. So, here it is. A employment, and am a pretty geeky per- a prepaid phone card - which is about short snip of Buzz’s message to me on son in trying new technology, but I am the same level of additional effort. Can Voice Mail. It’s not the same Skype qual- missing the attraction. Skype is great I eventually use a VoIP/ enabled cell ity; it does deliver the message. VMBuzz technology, implemented in a very novel phone to bypass my minutes allowance PodCast! way, with new features that are being yes, but the underlying wireless provider developed, but it is missing what I still will charge me for the bits through the [Snip] data connection. think is the marketplace. COOK Report: We find at: http://www. henshall.com/blog/archives/001106.html Journal VM Update February 1, 2005 05:58 PM Henshall: I’ve again been playing with Skype VM beta. I also wrote about it first here and then added these thoughts here. Today I terminated my little experiment with iPodRadio. I just wanted my iPod back. While it was neat getting visitors from all over the world listening to my music I wasn’t getting to use it myself. So it is now offline. Which brings me back to VM (voice messaging please!). I added voice messaging to my iPod Radio, thinking it would be a neat way What I want at home is a communications system that lets me make outgoing “calls” to people around the world, lets me receive incoming calls from anyone, lets me make emergency calls if I need to, allows conferencing, have voicemail (emailed to me), and isn’t expensive. I have solved that set of needs by using Lingo (VoIP) on my broadband connection, with no local line as backup, but with my mobile available if the Lingo/ broadband is out of action. In an emergency, I can use either Lingo or mobile whichever is closest to hand. The Lingo costs me $19.95 a month for unlimited US, Canada and Europe, which pretty much defines where I call. Per call pricing outside those areas is reasonable. 92 So - if long distance charges are increasingly bundled into VoIP (and non-VoIP) based service packages, and especially if this is bundled into my high speed internet/Entertainment package at home - why would I bother with Skype? Hope this doesn’t sound like the last wimperings of a fading telecom person! Stuart Henshall: I’ve not had time to read the history on the list. Just jumping in. Lingo is one of the best PoIP plans around. I think Broadvoice may be even cheaper for even more countries. As you note they are effective plans. The COOK Report on Internet May June 2005 However, your note doesn’t mention the impact of “presence”, ad-hoc conferencing, audio quality - stereo positioning is coming, IM/Chat Integration, video, and always on etc. Elements that are changing how we use and experience conversations. (In this realm there was also a nice post by Barlow recently called “The Intimate Planet” http://barlow.typepad. com/barlowfriendz/2005/01/the_intimate_pl.html . There is an “Online Presence Spiral” and this links to a great paper by Douglas Galbi. http://www.henshall.com/blog/archives/000718.html I’m a real advocate for intensive use. Not that Skype may be the only solution; it is elegant in its simplicity. There are fewer steps to make a call in Skype vs the phone or even most mobiles. It’s easier to return calls etc. That’s probably just hygiene factors. Where it really changed what I do is in “global” connections, sharing links while talking, and adding in screen sharing file sharing etc. The capability for an “independent” to hold a global conference call 4x per week for sometime hours really changes the “who” I can work with. Having audio where I can hear helps too. The majority of conversations I hold on Skype combine text and chat. That’s something I can’t do on the phone (Well I can with a Bluetooth headset and my Nokia using Agile Messenger and Opera, would be clumsy). Similarly, the “how” I contact someone is changing. It’s that “Presence” thing. In today’s world a “ring” may be rude, a text message intrusive, when a voice message may provide a timely update when you need it. The PoIP plans still ring, can’t text effectively, and don’t enable voice messaging. Still at the heart of it I’d choose a Skype call over a phone call any day of the week. It’s better sound quality. Until better audio comes along --- then I’d probably go there too. After all what I want to do is talk and listen. Retzer: Steve asked: I sometimes think I am really missing the excitement here in these recent emails about Skype , and I cant fully put my finger on what is missing. I’ve used it only briefly but it worked better than a popular VOIP service that charges. Only briefly because of the relative inconvenience of using a pc as a phone. Steve continued: I’ve then thought about using Skype from a PocketPC to avoid using a mobile ... Your comments didn’t mention that Pocket PCs increasingly have WiFi so that you don’t necessarily need to use the cellular time and you can overcome poor cell reception indoors if you have WiFi available. So, go on your business to a conference where you have WiFi. Connect your bluetooth headset and be fully connected all day with just the pocket PC rather than drag a laptop and disturbing folks when the phone goes off. Use your GSM time and cellphone service or Skype as desired - forward the cell to the Skype if you like. I’ve found a lot of conferences in places that have pretty poor indoor cell reception. Consequently, this is also a way around that problem. This is all speculative, however as I don’t know people doing this now. Forster: What excites me about is that they’ve taken an axe to a whole lot of infrastructure and associated people, allowing them to deploy a very high audio quality but extremely low cost infrastructure. You’re absolutely right that using a PC for a phone has some drawbacks. I believe it’s technically straightforward to put the client code into an dedicated device (an appliance if you will) so that a PC is not required. With respect to taking incoming calls off the PSTN: yes, that’s another very valid issue and huge limitation. But I believe that’s fixable as well. Heap: To prove I am not a caveman, I did call Stuart on Skype and had a good discussion about the merits of Skype in a business context - especially for consulting or global interactions. Quality was great! There are some good thoughts 93 below and I will watch the other comments with interest. Not convinced yet for “ordinary” residential use where I have already dropped Verizon and my entire “phone bill” is $20 a month, but I will watch this space! Ciscoʼs CMX Platform Promotes Mobility Across Wireless Access Networks Matson: Spent an afternoon with the mobile team of Cisco Systems earlier this week. They believe that the ‘mobile’ operators move into the ‘fixed’ space is the major trend ... and that providing mobile operators with a fully mobile IP ability - regardless of access network - is a key opportunity. But to hear Cisco talking about “creating network embedded intelligence” to counter the peer-to-peer IP connectivity across a “dumb network” was somewhat surprising! Forster: Hey, we’re a big place, and we try to sell to all sorts of different customers with quite different viewpoints. The groups responsible for selling to a particular set of customers tend to think like those customers. The groups selling Ethernet switches are lot more comfortable with the dumb network / smart edge notions, but even there the pain of dealing with the incredible mess of today’s desktop systems running Windows will drive many to desperate acts. Matson: They claim their Mobile Exchange (CMX) platform will permit the customers of mobile operators to roam over various mobile access networks ( GSM, 3G, WiFi etc etc). Not only that, because the CMX unpacks and categorizes every packet (to determine URL, location, file type, content, time, type of service etc.) it will afford mobile operators to support any business model - such as identifying a packet and either blocking it or charging for it!! I can’t believe Cisco has changed its core view of a world of “the dumb pipe with intelligence at the edge” if for no other COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA reason than Cisco makes intelligent devices and there is a bigger business down the line selling to the “edge” than “networks”. But in these chaotic, confusing and disruptive days transitioning into the IP world, Cisco cannot abandon the market potential and commercial opportunity afforded by major cash-rich players (like the mobile operators) who are “on the defensive” in the new paradigm shift and want to deploy intelligence to try and do anything they can to halt the Tsunami. Well now we know what Cisco are doing to help them “evolve into the future” (their words)! They claim to have 80 mobile operators ‘on board’ with their CMZ platform and will be making significant further announcements at the forthcoming GSM event. Cisco is desperate not to be relegated to commoditization and are even venturing with radical new charging model with this platform - e.g. $x per user. Worth watching Coluccio: “But to hear Cisco talking about “creating network embedded intelligence” to counter the peer-to-peer IP connectivity across a “dumb network” was somewhat surprising!” wrote Malcolm Matson. I agree that such a statement coming from a company that vowed to replace switched PBXes with its own VoIP gear sounds counterintuitive, but if their audience consists of fixed and mobile wireless operators, then I’d hardly call their statement surprising. Forster: Right. (And we did move off Lucent’s PBX’s, several years ago. Only recently we moved off the old voicemail system onto our own IP-based Unified Messaging System. It’s a big pain to move thousands of users to a different system). Coluccio: If a telephony model is based entirely on its intelligence residing within the client’s software there is not means of monetizing their services. It’s now axiomatic to say at this point that as application intelligence is pushed to the “outermost edge” (to differentiate from the ISP’s edge), and into the client’s end point software, so moves the value of that application. These circumstances are hardly the sort of heart-lifting prospects that vendors and operators who’d like to capitalize on the original dream of the shift to Layer 3 like to hear, much less promote, when addressing new markets with their wares. Matson: Don’t get me wrong. The longer we are in this period of paradigm transition (lengthened primarily in my view by misguided regulatory/public policy intervention) the more ‘visionary’ public companies like Cisco have to attend to the short-term interests of their shareholders by developing revenue generating opportunities created by the vested interests dedicated to extending their old-paradigm shelf life and slowing the transition. So I am not criticizing Cisco and I suppose I am not in that sense “surprised” that a major public company such as Cisco should seek to generate revenues by serving deep-pocketed vested interests rather than end users who will benefit from the demise of such cashcollecting intermediaries.... but then I suppose I was. Maybe it simply highlights the fact that once a public company - then the need to hit revenue/profit targets transcends all else - which is why, as an entrepreneur with private funding backing for me and the “vision”, I can be single-minded about where I am going ... trying to accelerate the transition of the chasm ... at whatever cost. I’m sure Bosack, Lerner and Co. know what I mean!! But What is the VoIP Market? COOK Report: Is VoIP now mature enough to be worth talking about only in the context of what the surmised customer base is? Does it make any sense to suggest that such a customer base falls into three broad categories? 1. Ordinary consumers - SYPE 2. Telcos - sip based proprietary systems 3. Enterprises - because of security issues they don’t like , BUT do they want the SIP gear being designed for the phone companies? If not why and what then do 94 they want? Coluccio: Narrowing the field of ‘customers’ down to just three, i.e., Consumers, telcos, and enterprises is in my opinion an over simplification. What happened to the large base of government-based networking taking place, not to mention an even larger base of enterprise customers, municipalities, greenfield operators, condominium builds, etc? To further break it down, some enterprise customers are asset-based network operators unto themselves (some in a very large way), meaning that they own major portions of their own infrastructures. I submit that in each of the categories of ‘customers’ listed above you will find both SIP and Skype infiltrating into traffic flows, whether they are sanctioned or not. Enterprise IT departments fill proxy roles similar to that which David refers to when he speaks about the vendor-telco, or vendor-consumer relationship. In the case of the enterprise, IT is both service provider and the vendor that counts, and has the last say in what is authorized and what is not. So, IT folks, too, have to listen to their ‘customers.’ And while there are many applications that make sense, like IM, those apps don’t always become endorsed or supported by IT in a timely manner. If they are compelling, they eventually find their way onto users’ nodes, even it if means putting the enterprise logo on them first. And this is exactly what I’ve seen happen in my client organizations. And in a way that we’ve seen this take place with IM already, where the Merrill Lynches (I believe), the JPMs and many others have implemented it, we’ll very likely see the same thing occurring with a -like application before long, if not licensed directly by and others, themselves. You asked earlier if I’d offer some examples of enterprise applications that might be conducive to , perhaps in banking or brokerage. How about junkyards? Junkyards and financial trading floors share more than one legacy together, you know :) The COOK Report on Internet May June 2005 One is a form of public address system application that is tied to a conference circuit arrangement known as a “hoot and holler” network, or, variously as, “shout down” circuits, or even “order wires” in some folks books. With some tweaking to ‘s conferencing features I can see where similar functionality could be achieved, although I don’t know just how far tampering with it would constitute creating another application altogether. See one vendor’s approach to hoot and holler over IP, below: http://www.bsslimited.com/Business/ KCS_IP_Hoot.html Cisco has done considerable work (as have others) in adapting this application to IP, as well: http://makeashorterlink.com/ ?B6EF5267A It doesn’t take too much imagination to see how ‘s conferencing capabilities might fit here. I wrote the following introduction to the article that follows, which should expose some of the reservations that I have, however, since force fitting Skype into enterprise will not be as acceptable as most would think. There are the usual security issues to deal with, but beyond those there are compliance mandates that must be honored with regard to disaster recovery preparedness, and Sarbanes-Oxley implications, as well. Remember, too, that many voicebased financial trading ‘circuits’ (oral session facilitators?) must be auditable to the same extent as their computer-data counterparts, meaning that they must be recorded (i.e., conversations are recorded) and tied to D/R backup facilities through a maze of failover gunk, besides, at least to the extent of meeting minimal criticality measures. What follows is a message that I posted to my SI Forum earlier today, prefacing an alwayson.com blog post. I’ll play it back here in its entirety: --[Frank Coluccio: After having used Skype for only about a month, not to mention having read several very recent in-depth interviews with its founders, I have found all of the implications of that are offered in the article below to be at least highly plausible, and most of them true. What the author states about the RBOCs resolve to see their convergence plans to the end this time as opposed to their earlier announced plans of ten years ago is particularly in line with my way of thinking, since for them NOT to follow through with robust video capabilities would amount to their hanging up their sneakers for the last time. http://www.alwayson-network.com/ comments.php?id=8478_0_4_0_C ] Begin "alwayson:" The -TV Connection VoIP is forcing telcos to go Hollywood. Michael Stroud [iHollywood Forum] | Posted Feb 9: Since Skype invaded my life, I understand why the telcos are so desperate to take over my TV set. My kids’ nanny makes one-hour phone calls for free every day to her family in Hungary. I myself plan to call my family and office in Los Angeles for free when I’m in France next week. And a buddy I told about Skype a week ago is already calling a Canadian business partner every day, for free. Worldwide, about 67 million people have downloaded Skype. How long until one billion people realize that they can make unlimited free calls to their friends? How long until everybody does? One, two years? Until then, I can still switch my entire office over to a Voice-over-IP service like Vonage, and save around $500 a month on my employees’ phone bills—and while I’m at it, port my phone number and incoming calls to my laptop at Starbucks. Hell, I could even Skype over Verizon or Sprint’s EVDO networks—getting unlimited broadband Internet AND voice calls for around $80 a month. So what’s all that got to do with TV? Everything. There’s no margin in voice calls any more, and the telcos know it. If they want to survive, they’ve got two options: bundle commodity services together (wireless and wireline voice and 95 broadband internet); and start zapping lots of video down their fat pipes. In other words, get into the TV business. It hasn’t been lost on the Bells that while the fees they command for voice calls have been collapsing, cable companies have been raising their programming fees, year after year. That’s because— satellite notwithstanding—cable operators have remained monopolies in their markets, while the telecommunications market has exploded with competition. This is a qualitatively different situation than the telcos faced ten years ago, when they announced expensive and impractical trials to bring video-on-demand, interactive TV, and other services to homes. Or when Hollywood super-agent Michael Ovitz contracted with Nynex, Bell Atlantic, and SBC in 1994 to develop interactive services over phone lines. Then, the telcos were dilettantes with deep pockets. Today, it’s succeed or die. You can still be skeptical about Verizon’s and SBC’s publicly announced plans to invest billions of dollars to offer IPTV to consumers. But don’t doubt that this time they will stick to their TV plans to the bitter end. If they fail, they will either buy a cable company or be bought themselves. After all, ten years ago, who could have imagined that mighty AT&T’s long-distance business could go for a paltry $16 billion? In their battle to implement IPTV, the telcos’ advantage over cable companies is the same as their disadvantage: they’re starting from scratch. “The telcos are in the enviable position today that they don’t have this massive legacy investment in all this cable video technology like cable companies do,” said Ed Graczyk, marketing and communications director for Microsoft Television, which is providing the underlying software for SBC’s and Verizon’s IPTV services. Cable companies, Graczyk said, have “billions invested in the underlying technology and services.” Cable networks are designed primarily for programming; broadband internet was added later and uses a different COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA architecture. Because interactivity was added as an afterthought, uplink speeds are slow. SBC and Verizon plan an integrated, open IPTV architecture that will offer lightning-fast user interaction with the network and nearly limitless channel possibilities. Think instant channel changing (no one- or two-second delays after you click the remote control), thousands of on-demand channels, multiple picture-inpictures, and multiple camera angles. Cable companies won’t take this lying down. They’re planning to ultimately upgrade their own networks to IPTV, although they’re tapped out from their massive investments to go digital. For both sides, the Holy Grail is offering so-called Triple Plays or Grand Slams: packages of three or four services that will reduce customer churn and create the economies of scale they need to profit in an intensely competitive environment. For Verizon, that means offering wireless and wired voice services, broadband Internet, and television. For Comcast and Cox, it means adding telephone service to their cable and broadband Internet services. For both cable operators and telcos, telephone service will be the commodity product and perhaps even the loss-leader they need to drive consumers to more profitable multimedia services. Because if they try to gouge you for your phone calls, you can be sure they’ll turn you into a Skyper, too. The Importance of Directory Services Stastny: Regarding Gordon’s three categories [of broadband], - I would put municipal networks under enterprises, as they have essentially the same requirements The 3rd point can be split in two: a. Enterprises connected to telcos b. Enterprises wanting to peer directly Case a is supported within the sipconnect group that is found at http://www.sipconnect.info/mc/page.do with its interconnect draft in turn found at http://web. memberclicks.com/mcdatafiles/site/sip/ SIPconnect_Version_1_Draft_2-2-2005. pdf Case b may the most important use case for ENUM Fixed Telcos and mobile operators are currently coming back to the market of VoIP very strongly, both in the US and in the rest of the worlds (see 3GPP/TISPAN/ATIS using IMS, but finally the whole issue boils down to the question: will the consumer need telcos anymore or not? Note that the SIP server manufacturers play a crucial role here, they may even be betting on the wrong horse. The horses they should be attending to is the device and end-user application providers. Device manufacturers basically do not care who is buying their equipment, just look at the mobile phone industry Of course they like to sell in bulk via service providers, but they also sell to the end-users. And any general-purpose mobile phone or GSM-enabled PDA is another step to the death of the service providers. The big US carriers, the big mobile operator groups and also the cable operators are currently competing like mad within and between the groups and they seem completely forget Clay Shirkys’ Zapmail example: http://shirky.com/writings/zapmail.html You can compete with everybody except with your customer So this battle will be decided from the customer, and the first one really satisfying the customer’s needs will succeed Skype is well ahead here Coluccio: Hi Richard. I had to read your message twice before I realized that you would have been much better served in my opinion if you had prefaced it with the following qualifier: “From a directory services perspective, ... “ Stastny: Hi Frank. Sorry, I am currently 96 so immersed in addressing, numbering and naming issues that I assume automatically everybody else is also ;-) Coluccio: There’s hardly any cause to sound an apologetic tone. The single most rewarding aspect of participating on this list, if not challenging at times, as well, has been the kaleidoscopic effect it’s had for me and I’m sure others, seeing the world through so many other professionals’ eyes on a sustained basis. I certainly appreciate your viewpoint and those of others, even when connecting the dots, using both real and virtual links, sometimes requires cracking a dust collector, or two ;) Stastny: In my opinion what will be left in future as service on the Internet, besides hosting services will be “directory services” or “identification” services. This starts with the existing DNS, continues to the move of existing number translation services to the Internet (Number portability, IN services), using ENUM or something else, RADIUS etc. for roaming, certificates and public keys, and will end with RFIDs. Coluccio: Failing that measure [directory services], your message would have end nodes connecting to one another through an imaginary ether. It appears that you’ve entirely ignored both the network and physical layer dependencies below the services layers in your presentation. How do enterprises connect to one other independent of telcos or cablecos or WISPs and ISPs, for example, regardless of what form of addressing is used? Stastny: This is easy: the “imaginary ether” is the IP network. It all boils down that you know the IP address of the other party. Ok, some additional information is helpfull, but this also can be served with the existing infrastructure, the DNS und URIs To get the other “ether”, the E.164 numbers in, you need a translation service, e.g ENUM. So what you need is a namespace and a centralized service (with the current development in P2P you may not even need this) The COOK Report on Internet May June 2005 Regarding IMS, this discussion here is a bit US centric and IMS was up to now more Europe centric. I personally am not convinced of its success (in my opinion it is much too complicated), but mobile operators like it and now also fixed operators (TISPAN and ATIS. End of March there will be a joint 3GPP/TISPAN/ATIS workshop on NGN IMS in Washington. Earlier - Coluccio: This raises yet another question: Can a discussion concerning the manner in which naming and addressing is accomplished be fully meaningful without also taking into account the service organizations that are supporting access, aggregation and transport? I suspect that the answer to this would be yes, if one were looking solely at the services layer and ignoring everything beneath it. But if you zoomed in to a set of real world conditions, would it? I’m not looking to nit on this. I’m merely highlighting my view that one must be mindful of all adjoining layers when addressing solutions to any one of them, if they are to be held responsible for managing the services of any of them. On another note, unless I missed a portion of the discussion here, yours was the first message I’ve read on this list, aside from an article of my own several weeks ago, that even mentions the acronym IMS, by which I take it you mean IP Multimedia Subsystem. I find this a bit curious, given the nature of the discussions that have been taking place here. I’ve surmised that there must be a natural aversion to it, which is fine in one’s own mind, except that by not exposing the elements of IMS, and its potential implications, I find the overall discussion wanting to some degree, due to the inevitability of having to contend with its existence in the future. Skype and Motorola Marketing Partnership & Motorola Ultra Cheap Mobile Coluccio on February 14: I thought it a bit odd that there would be no mention of this release here on the list. There doesn’t appear to be any mention of it on their web sites yet, either. If anyone can shed any additional light on this, it would be welcome news. From Converge Digest: 15-February-2005 Volume 12, Number 029 “Motorola and Technologies announced a co-marketing collaboration that will provide greater connectivity options and access for Skype ‘s more than 25 million registered worldwide users. The alliance will explore opportunities broadly across both companies. The initial focus of the collaboration will be on co-marketing of new optimized Motorola ‘ Ready’ companion products, such as Bluetooth headsets, dongles, and speakerphones, as well as delivery of the Skype Internet Telephony experience on select Motorola mobile devices. Motorola ‘ Ready’ companion products are expected to be available in the first half of 2005.” COOK Report: I knew about the announcement but only because I had a Chat with Stuart Henshel who informed me rgarding it. Stuart told me that Dmitry Goroshevskii is doing a federated peer to peer challenger to Skype. Would that be good for use by hoot and holler systems I asked? "Ohhh yes," said Stuart. Coluccio: Say more about Goroshevskii and federated, please. Henshall: Martin Geddes wrote a piece on [Goroshevsky's] Popular Telephony and Peerio that may be interesting to you. Generally the jury is still out. Are they vapor or not. I’m yet to see anything ever. Still they get press, I don’t know why. The ideas are interesting. See http://www.telepocalypse.net/archives/000602.html “Telepocalypse to Goroshevsky: So Skypes‘s problem is a lack of private namespaces? Dmitry Goroshevsky: Exactly. In our system you run your own Peerio network, and join other Peerio networks as a special administration task. You authorize a handset to join another network, from which it receives an identifier, and it can receive calls from the other network. It’s modular. So just as an example, if you have another company that you work 97 with a lot, you can create another set of numbers you use just to communicate between your two companies.” Henshall: Now that’s a “federated” approach. Note not confederated. It also has interesting security implications. Separately I found myself playing around with a crude number today. Skype does provide #of concurrent active users and “number of minutes served. Today they broke 2 million concurrent users and did over 3 million minutes in that hour. The “Skype” time experience is like no other communications experience to date. In my view the above confirms that. A possible hypothetical best guess average usage around 500 minutes per week. I played around with them here. http://www.henshall.com/blog/archives/001120.html I’ve also been mulling over the Moto deal with Skype. The minute’s case above means the “profile” of a Skyper is not one that I’d think fits easily or scales within the current mobile package for voice or data. Thus with handset manufacturers it may present a real opportunity. Late night thinking is found at: http://www.henshall.com/blog/archives/001121.html. Could Skype’s real “case” already be in the numbers? Could it be that Skype doesn’t really know themselves? Quite possible. Matson: My inside contact in the London Skype team told me some weeks ago that they will be launching a branded WiFi handset for the Christmas 05 market - so we now know who the mfg. Is [Motorola] ...and what will be top-selling gadget next Christmas! Peter Ecclesine: “Operators develop ultra-cheap mobile” By FT.COM – Feb 15 “An ultra-cheap mobile telephone has been developed by an international consortium of network operators, in a move that could transform the market by bringing mobile telephony within reach of millions of people in developing countries. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Motorola, the world’s second biggest mobile phone manufacturer, has developed the C117 handset in partnership with the GSM Association, which represents operators worldwide, to overcome the principal barrier to mass phone ownership: price. The low-cost mobile project was initiated by nine operators, mainly Asian, and the scheme will be piloted in Asia. The GSMA has set an initial target of 6m handset sales in the first six months. With other manufacturers expected to follow Motorola’s lead, operators hope the scheme will ultimately drive the global industry towards 2bn connections. There are almost 1.5bn people connected on mobile networks worldwide. The C117, which is expected to be the first of a family of low-cost phones produced by Motorola, is likely to be sold for less than $40 at wholesale prices less than half the $100 that the GSMA defines as low cost. The GSMA, which hosts the industry’s biggest trade show in Cannes this week, estimates the C117 will open mobile phone ownership to more than 700m new consumers in Asia, Latin America, the Middle East and Africa if it can ultimately bring the price below $30. Rob Conway, chief executive of the GSMA, said that while a low-cost handset offered the prospect of new growth markets to operators and manufacturers, it could also aid social and economic development in emerging markets. “This is a stimulus that can really kickstart the economic development of certain areas. People get connected and that helps them to build up their social and business infrastructure,” he said.” [Editor’s Comment: This is just one more piece of the emerging jigsaw puzzle. It calls to mind the points so forcefully made by Sam Pitroda and CK Prahalad in May of last year and published in our September October 2004 issue. Namely Asian markets are huge and consumer electronics development for those markets will need to deliver goods that are very different from the US and European Markets. I suspect that this cell phone development is a very significant move. Increasingly there is a very viable wireless infrastructure in all parts of India and China – imagine running on 100 million of these new very low cost cell phones in two years time.] Goroshevsky, Popular Telephony and Peerio COOK Report: Thanks to Stuart for his pointer above to Martin Geddes’ Interview with Dmitry Goroshevsky. I have read it and find the concept very intriguing. A couple of questions. Would any self-respecting enterprise ever install code like this without being able to inspect every line of it? Is the guy going to claim that his stuff doesn’t have to go through NATs because its is used with “federation” and can use his GNUP whatever as a gateway to get where he needs to go? Coluccio: Gordon, it may have slipped your attention, but we discussed an enterprise adaptation of the other day very similar to this during our own call. You may recall my mentioning WalMart, by name, supporting its own p2p voice applications with restricted access controls, and using gateways [GNUP? in this case?] to its suppliers and to the PSTN, where ‘WalMart’ permits ;-) In generalizations on this topic, I don’t think that I ever, or at worst I have seldom, referred to the calling model introduced by Skype in a singular sense, always preferring instead to use verbiage like “Skype-like,” or “and other similar applications,” for this very reason. COOK Report: Is no one anywhere really USING Goroshevsky’s stuff? Richard Stastny: I had a entry (rant) about this on my blog in November 2004, http://voipandenum.blogspot. com/2004/11/gnup-numbers-anotherhot-air-teaser.html I copy it in here for convenience: GNUP Numbers - another hot air teaser from Popular Telephony 98 I am not an expert in marketing, but one potential strategy seems to be to get all potential customers feed up with the company before the service really starts. Whereas the competition (Skype) is announcing new services only after deployment, “Populistic” Telephony is only deploying hot air. They also seem to have a weird understanding of the meaning of weeks and days. In my simple understanding a week is a week and a day is a day. Not so with Peerio, GNUP, PT or whatchumacallit. Since May 2004 they announced on their Peerio444 <http://www.perio444.com> webpage that new beta-testers will be accepted within two weeks. In the meantime the webpage was replaced with “looking forward” into the void. [Editor: On March 24, 2005 - the above url was no longer valid.] At the Fall VON 2004 they announced at the both, the service will start November 1st, 2004 and will be called GNUP. OK, on November 3rd one could download the so-called PT Inspector (6MB), which did basically nothing but check if something is available at the PT web-site and tells you the service will start in three (3) days. After one week it stopped working (did not find the PT web-page, so I downloaded the new version (only 4MB now), containing a patch to find the webpage again - and obviously a optimized software saving 2 MB to do this. Then the next teaser was announced on the blogs: you now can get a number! So I retrieved my GNUP number: (8844) 4294967265 Since Tom Keating <http://blog.tmcnet. com/blog/tom-keating/voip/voip-blog/ popular-telephony-gnup-registry-nowlive.asp> got (8844) 4294967281 registering a bit earlier, they seem to count down from 4294967299 ;-).Who came up with this number? A random generator? Or has this some secret numero-logic meaning? Freemasonry? The PT Inspector still tells me the service The COOK Report on Internet May June 2005 will be available in three (3) days, so somewhere in 2006, maybe. Anyway, I wonder what (8844) means. From talking to the PT guy at the VON and reading the flyer they distributed, they are planning to make a petition? PT to ITU-T to get country code 884 (which is currently reserved) assigned for their use. Although I told them that you do not get CCs from ITU-T assigned by petition, but by request and that there is no way to get a CC assigned to a carrier (here you may request one out of 882 xx - which is easy and does not take much time - assuming a proper request), but only for a service., and here you need first a service description. This path is more time consuming (as I know from personal experience ;-). On the other hand, they could either use 878 10 or they could even request 878 20 or so. But they seem to know better and just start to use 884, because it is unused. ITU-T will really like this. I wonder which measures will be taken eventually by national regulating authority if carriers within their reach start to route numbers on the PSTN to non-existing Country Codes, considering Resolution 20 <http://www.itu.int/ITU-T/wtsa/ resolutions04/Res20E.pdf> of the WTSA 2004 <http://www.itu.int/ITU-T/wtsa04/index.asp> , which instructs: 4. the Director of TSB, in close collaboration with Study Group 2, and any other relevant study groups, to follow up on the misuse of any numbering, naming, addressing and identification resources and inform the Council accordingly; 5. Study Group 2 to study, urgently, necessary action to ensure that the sovereignty of ITU Member States with regard to country code numbering, naming, addressing and identification plans is fully maintained, as enshrined in Recommendation E.164 and other relevant Recommendations; this shall cover ways and means to address and counter any misuse of any numbering, naming, addressing and identification resources, and of call progress tones and signals, through proper development of a proposed resolution and/or the development and adoption of a Recommendation towards this aim. Enterprise Voice Issues Yield Many UnKnowns COOK Report: I am left with the impression - a rather bizarre one at that - that the consumer space with Skype, Vonage, Lingo and all that lot is seriously far ahead of the enterprise. Why? Forster: What’s so strange about that? Consumers, or at least early adaptor consumers, can try out new things much more easily than large enterprises. Enterprises would inevitably go through an evaluation process and would uncover 50 different reasons why anything new is incompatible, insecure, unreliable, or just too new to adapt so they have to stick with the status quo. At this stage in the market for new telephony services the new players can claim success with 1-10% adoption, but no enterprise IT department would endorse something unless they expect to use it for most or all employees. IT departments like uniformity not diversity. Eventually, if there are large cost savings or productivity increases then the enterprises will respond. Shockey: As per other comments its not surprising that consumers are pushing VoIP ahead much faster than enterprises, With Skype the benefits are simple, demonstrable and immediate and they don’t have to go to committee to define a implementation schedule. COOK Report: Are primary issues there ones of security and interoperatbility? Forster: No. A primary issue is decision process time, i.e. inertia. Shockey: Well yes and no. The security mavens in many enterprises have demanded that Session Border Controllers (aka fire walls on steroids) be installed to present a single point of ingress and egress to the internal VoIP network and the effect of these devices are not completely understood by anyone including the vast majority of the SIP community. Some Edge Network Element was necessary to implement STUN ICE etc but 99 some vendors have made promises for these devices I find hard to accept. Service providers think SBC are necessary since they need a point of ingress where they can maintain and meter state for calls that must terminate on the PSTN and do things like “SIP mediation” which is basically fix implementations that may or may not be standards compliant. In addition the security model for SIP trunking is not well developed. it is assumed TLS between proxies but this does not solve the termination carriers problem of billing which means MCI need to know with some level of certainty that joe.blow@ibm.com who is trying to create a session is really Joe Blow of IBM. Jennings: I think this is somewhat wrong. RFC 3325 provided a solution to this and in truth MCI does not care if it is Joe or not. They care they can prove it was IBM and that IBM will pay the bill. MCI offers this service today so I think it is pretty hard to argue something is missing. COOK Report: Where does Sarbanes Oxley fit into all this? Does it mean that there are whole new escalating concerns of security that mean that a vendor that sells security has a big edge over one with just cost on its side.... Shockey: In financial industry applications this is a big big deal. Reuters for instance developed a private secure SIP based IM system for bond traders. $18K per seat per year. COOK Report: Are enterprises really shy about innovative VoIP technology because they don’t have interoperability and because they are scared of proliferating security dangers? Forster: They’re probably concerned about spending a bunch of money on staff and equipment and then running into bugs and delays, and not delivering results. Henshall: Suggestion: Who is leading the revolution in call centers? I’m not referring to the back end rather re-think- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA ing how we that is you and I engage with an 800 pound gorilla positioned to win ages do carriers have to offer enterprises a call center. At the moment there is lots the enterprise? What kind of interoper- that will satisfy combined concerns of of push one or two, wait and hold. Then ability is needed that isn’t yet here? What cost and these other issues? you get dropped. Tried a HMO recently? kind of security? Shockey: Well remember what is the Hung on for half an hour. The paradigm that is emerging is a world in which we Shockey: Its not a security model prob- Holy Grail here. Direct SIP trunking of have no need to hang on. We call, click lem. It’s a trust - identity management calls over IP from the edge of the Ena call to: tag on a website, get a text problem. How do I know you are who terprise directly into a Carrier Network for termination. Eliminate or reduce the message or a simple voice recorder for you say you are? number TDM T1 trunks at the edge of capturing our message, agree to share our “availability presence info short-term, Jennings: Well as Richard knows I’m the enterprise and you tip the economic get an update in terms of likely return one of the author on the IETF solution balance way in favor of VoIP and mess call time and then get the right operator to that problem and one of Richard’s up the LEC’s big time. calling us back when it is still convenient colleagues, Jon Peterson, is the other cofor us. We’re less frustrated. The opera- author (see http://www.softarmor.com/ COOK Report: Is the enterprise really a tor is better selected. Plus there are some wgdb/docs/draft-ietf-sip-identity-03.txt ) mess from the standpoint of voice as an application on its networks? If so what compelling additional things the co could However, I’m not sure I see this as a steps must be accomplished to improve do as a result. problem. You have no way to tell whom things? Move from caller on hold to call back re- this email is from. If I called you on quests combined with text support where the PSTN, you would have no way to Shockey: It’s lots of things. Remempossible. Share short-term presence info know that my caller id was correct. That ber SIP and ENUM like technologies to facilitate the connection and recon- works fine for enterprises today. I agree completely decompose the features and nection if necessary. Etc. For web based that knowing who you are would be an functionality of the PBX into plug and catalog businesses I’d think it would be improvement to communications - but I play components You are beginning to immediately compelling. Similarly for question the idea that lack of it is a show see the emergence of demands by major enterprises that - gee, I like those Cisco Banks and online Financial institutions. stopper. phones but I want them to work with my For not much in monetary early learning Don’t get me wrong I think knowing Avaya IP PBX. Oh I’ve got 7 different could be captured now. who I was communicating with would Voice platforms ..I want to manage a unibe a big step forward - I just don’t think fied dial plan across the enterprise that COOK Report: How can small vendors that it is table stake because no one has all of the platforms will support. compete under these circumstances? it today. Jennings: All the open source PBX Shockey: You’ve never actually had to What happens as Skype goes on mobile (Vocal, SER, Asterix, sipX, etc) seem to configure a Cisco Call manager have devices? How does the enterprise protect work great with Cisco phones and gateyou? They have lost several major stra- itself from voice conversations it can’t ways. I asked Avaya if their PBX would work with Cisco phones but so far they tegic accounts in the past couple of years control? have not got back to me yet :-) -- including Merril Lynch because of the inherent problems and lack of native SIP Forster: Look at what’s happening with IM’s, PDAs, and cell phones. Enterprise Forster: If things look like a mess then support in Call Manager IT departments in some cases try to we’ve passed out of the early euphoria Jennings: Funny you should mention control these but I suspect in many cases stage, which is good. We’ll just have this. We did loose the Merrill deal some- their employees start using there before to keep muddling along and see what works best and who does it best. There what over a year ago however I never they’re ‘supported’ by IT. is huge change taking place, but if we’re heard SIP mentioned as something that was involved in that decision at all. How- Shockey: Like DUH, Skype simply disappointed by the short-term expectation gap we may still be amazed with the ever, Merrill after doing some work with works. longer-term results. the company we lost to reversed their decision and decided to go with a Cisco Jennings: What happens when you use system. Needless to say we were pleased AOL, Yahoo, etc inside lots of enter- Henshall: Telio www.teleo.com was preprises. The Firewall blocks it. Some sented at Demo@15 this week. It’s SIP - The press release is at enterprises will choose to block Skype and uses the GIPS voice engine. They http://newsroom.cisco.com/dlls/2005/ too. The owner of the network will assert have US numbers available. Voice qualprod_020905b.html policy on how it is used. Usual cat and ity is comparable to Skype in all ways. The client is effectively still PoIP. It’s mouse games will ensue. not Peer to Peer and all calls are routed COOK Report: How can Nortel and Avaya compete with Cisco? Is Cisco then COOK Report: What kind of VoIP pack- through the server in SF so it won’t scale 100 The COOK Report on Internet May June 2005 the same way. It has no chat facility and presence is only online or offline. It’s a good Vonage replacement today. Fee and prepaid minutes. Minute charges appear very competitive. Coluccio: note what this article says: ‘... an alternative to Microsoft Windowsbased machines, can sell for as little as $200.” (It appears that PCs, like voice, are in a race to the bottom. Where zero wins? C’mon, now... how far can you leverage free... I suppose we’re getting ready to see.) http://www.internetnews.com/infra/article.php/3483726 Skype, Xandros Bundle VoIP, Linux By Colin C. Haley February 16, 2005 BOSTON -- Internet telephony specialist and desktop Linux developer Xandros will bundle their products and sell them through retailers including Amazon.com and Walmart.com. The agreement was announced here Wednesday at the LinuxWorld trade show in conjunction with a preview of Xandros’ new Surfside Linux operating system for consumers. “We share the same values in trying to lower costs for our users,” Eileen Broch, director of product management, said, adding that the agreement marks the company’s most aggressive move in the Linux community to date. With more than 25 million registered users, Skype is surging. Its software, which uses an instant-messaging-like interface, enables free Voice over IP (define) calling between its broadband users. The Luxembourg-based company, which is privately held and venture-backed by Draper Fisher Jurvetson and others, wants to boost revenues with paid offerings, as well. As part of its pact with Xandros, users will receive credit for about 120 minutes for the company’s Out product, which enables calls to landlines and mobile phones. more powerful and broad generalization here that can be drawn here. Regularly, the Out service costs about 2 cents per minute for calls to 20 of the world’s largest markets. Other paid services, such as voicemail, are expected soon from Skype. Skype is not the first Internet application that has become viral, and it would seem that the keys to becoming viral are: Skype officials believe today’s deal will expand its base. Xandros’ Surfside OS also includes the Firefox browser, Thunderbird e-mail, a headset and security, including anti-virus and firewall applications. Xandros-based computers, an alternative to Microsoft Windows-based machines, can sell for as little as $200. Financial terms of the -Xandros pact were not disclosed. The pact comes only a day after Skype trumpeted a broad agreement with mobile phone giant Motorola (Quote, Chart) at the 3GSM World Congress in Cannes, France. That partnership will focus on collaboration and joint marketing of “Skypeready” products, such as headsets and speaker phones, as well as lay the groundwork for delivery of service on Motorola mobile devices. Skype Not Tied to Specific Hardware Can Act as a Virally Infectious Communications Agent COOK Report on February 18: Let me see if I can articulate a couple of threads after having talked with Stuart. Richard Shockey was saying to me and others more than two years ago that voice with VoIP is just a bucket of bits. YES - it its the SOFTWARE “stupid.” – Yes, he called that one right. But things have moved on since then big time. What I think we may be seeing here is do the basic software voice program right, then let it lose and watch it virally infect everything. Tony Li: I think that there is a much 101 - Provides significant practical value Broadly deployable - Free Moreover, the opinions of those stuck in the previous paradigm can safely be ignored. Broadband (and more generally -- higher bandwidth) always enables new applications and since IP is the ubiquitous transport, the only real delay is in discovering needs and engineering the solutions. This is the real message of the Internet: communications is only limited by your imagination. Complexity is evil, and pragmatic considerations rule the day. Bandwidth uber alles, COOK Report: BUT for the infectious agent to do its work it cannot be JUST a Wintel platform. It has to run on EVERYTHING Which is another way of saying that the software has to be severed from the hardware. Ergo it is another way of saying WHY Skype runs on Windows, Linux, Mac OS-X and if I recall correctly soon on Symbian for smart PDAs that act as phones, and on the Microsoft agent that works on cell phones? Forgive my lapse of knowing the proper terminology. You know I am Neaderthal in many respects -- no I-pod, no cell phone, never owned a PDA, so I am being drawn thankfully here into some new areas. Not having these other devices (cell phone, PDA etc) to play with has meant that I have been slow on the uptake. Anyway back to the main subject at hand - while I understand corporate concerns and security concerns thanks to Frank, Melissa and others I am concluding that the profound story here in "VoIP land" is not Vonage and Ciscos stuff, and SIP, and IP pbxs etc - this stuff, one way or another is tied to hardware and not surprisingly tied to the phone system and its related hardware. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Instead, this stuff (Skype), free and viral, is tied only to one thing. To BROADBAND and its underlying infrastructure. It could care less whether it rides over DSL or cable modem. It does both with equal ease. And now it is being freed to ride on mobile. Namely on Wi-fi and cell phones. Skype certainly depends on all the existing layers of the protocol stack. In that sense there is nothing magical about it. In depending on broadband it has to depend on the infrastructure that enables broadband. While end user premises and connections would need some serious upgrades to run end user controlled ligthwaves, these same end user connections, if they deliver broadbrand, can deliver Skype without any changes to the underlying network infrastructure. Skype will move forward in the rest of the world whether the enterprise likes it or not, and the applications that will grow up around Skype will give broadband another huge push that just email and the web cannot. Sebastian Hassinger: I believe that Skype is just another sign of the inevitable restructuring of the communications sector from a vertical to a horizontal orientation. Whereas the landscape has been historically been dominated by vertically integrated firms who own and operate networks, selling bundled connectivity and ‘value-added services’ to its subscribers, we are moving towards horizontally layered network providers (various flavors of broadband, wireless or wired) with the ‘value-added services’ (applications) provided by unaligned firms that ‘float’ on top of the networks. This is a natural evolution for maturing networks, and in fact one could argue the communication sector’s transformation is way overdue, retarded by the incumbents’ reluctance to change and the regulatory deck that has been stacked so decidedly in their favor. One reason this decoupling is inevitable is the economic reality that the optimal business models for selling network connectivity and network applications are at odds with one another. An operator wants to drive as much high value traffic as possible across its network and an application provider wants to reach the broadest audience possible with its service. Therefore, a vertically integrated operator/provider is always internally conflicted - the network business wants its value-add services to be bound to the network it owns, in order to increase the traffic and the value of that traffic. The value-add service (application) business, however, doesn’t want to be bound to the network because that limits its market reach. Imagine a -like application owned by a DSL or cable provider that favored its own network - allowing free calls only to subscribers on the same network, for example, and charging for calls that ‘leave’ the network. Imagine this network-bound -like app competing against the actual Skype with its polymorphously perverse disregard for the underlying network of its users. No contest. In fact this is the often the case with the features of the mobile networks today - free calls to other subscribers on the same network, or Push-to-talk only interoperable with phones on the same network, or even the spotty operation of SMS across networks. There is a robust business in operating ‘commodity’ networks that are agnostic to the applications that push bits across them, but it is a far leaner model than that practiced by the vertically integrated OpEx heavy firms of today. Some will make the transition gracefully, some won’t. Matson: Precisely, Gordon - you are getting hold of the OPLAN principle coming at it from the other direction. It is software that makes the link between networks - not networks concluding interconnection agreements in the old world way. As I have explained before - the OPLAN “island of local connectivity” is the essential building block - totally unconcerned and uncontrolled as to how and where ‘bits’ enter and exit the OPLAN -- unlike a central-office-centric cableTV or telco network. Coluccio: Don’t forget that with Skype, in and of itself, is not *the* content that 102 is being delivered. It is, in fact, a part of the underlying stack that enables delivery of that content, instead. Why am I nit’in on this point the way I am? I’m sensing, not only here but elsewhere, as well, that some would have regarded Skype as some form of ethereal manifestation poised to be used as a wholesale substitution of all that has preceded it, when, in fact, it merely muscles in where it can in the existing structure, becomes a participant in that structure, and then steps onto the shoulders of existing constructs in a very selective and focused way, instead, so as to be unencumbered by their usual constraints. In this way it is able to ignore much of the gunk and glue that make national and international networks function the way that they currently do. Skype is not the message. It is a part of the medium. Skype from an Enterprise Point of View COOK Report: Melissa Davis has been watching these discussions from the private enterprise, proprietary, intellectual property, security point of view. The universe she lives in is drastically different from the type of world that Stuart and I were talking about in our interview. She sent me some of her thoughts privately. We talked at length by phone. As a result I would like to open a new thread that will talk about the reasons that a very significant segment of enterprises will never allow Skype inside. Why they could because of Skype’s architecture never afford to allow it inside, even if it were the enterprise’s own virgin version of Skype run independently inside the enterprise’s firewall, I did get the impression that enterprises might begin to build their own Skype like applications that would do Skype like things in a way that the enterprise felt secure in being able to meet its legal and financial responsibilities in a way it could not do with the current version of open but proprietary Skype. There are lots of questions here. Among The COOK Report on Internet May June 2005 them are whether the big enterprises feeling breathing at their doorways will create their own like applications sooner than otherwise would be the case? will these parallel forces enable the telco’s to trash their circuit switched networks sooner than otherwise would have been the case? Stastny: E.g. SIP? There is basically nothing you cannot do in SIP what you can do in (even P2P), you may even do more and what is even more important for enterprises: it is not proprietary and you know exactly what is going on inside. Enterprises hate proprietary solutions, they do not want single source. The big advantage of Skype is that it is simple to use, but this is not a big argument for an enterprise. It is not between ENUM and Skype, it is SIP vs. ,Skype but you are correct somehow: for ENUM you need URIs, basically sip URIs. There is nothing like a Skype URI. Consequently, Skype is a walled garden NGN, more walled then the 3GPP/TISPAN/ATIS NGNs. Consequently, the decision will be made by the enterprises: if they will go SIP (and ENUM), the hype will fade away, because the normal users will use the same client. But if the companies are not fast enough to provide their employees decent communications and simple SIP clients to use on their laptop, PDA and Smartphone, the employees will continue to use Skype. Skype is now used already by many business people on the road. COOK Report: Are we indeed headed to a world where the only thing monetizable will be bandwidth? Forster; No. While I agree that Skype, and VoIP in general, are adding to the pricing pressure on voice-type services, I don’t think bandwidth will be the only thing that will be monetized. Whenever an amorphous and ill-defined service that is done somewhat differently for many sets of customers at great expense can be crystallized into one thing that can do 90% of what everyone wants very conveniently and at low cost there is a great opportunity for profit. What seems to be different is that for telecom the service used to be defined in terms of what came out of the wire (telco demarcs), or how the device behaved (cell phones, IM devices, etc.), but now the only constant is the IP packet and IP address; all the rest is in constant motion and the result is a lot more like software. Adding to the confusion, sometimes the software is momentarily solidified into devices (Skype in a phone). Coluccio: Jim, one word in your message caused me to reflect: “momentarily”. Unless it was merely extraneous verbiage, please explain why you chose to qualify it as such. I’m tending to think it’s got something to do with the limited capacity of some mobile devices, but not sure. Hassinger: This may be my personal prejudice warping what you wrote, if so, please correct me. As a related aside - if we are headed for a bifurcation of the providers of applications/value-add services from those of commodity connectivity, who will subsidize the purchase of the sophisticated device for the end user? Forster: Hmmm. I don’t know if the bifurcation will be so strong as to preclude bundling deals between the various players on either side of the connectivity/application boundary. Or even investments by the existing providers in the newcomers, which of course would facilitate bundling. COOK Report: Lots of unanswered questions.... Melissa? Frank? Forster: Frank, what I was thinking was that the functionality is somewhat frozen for a time so it can act like a fixed-function device, but recognizing that it’s really software and need not be frozen forever. Skype and Enterprise Security Issues Hassinger: Jim, am I right in assuming that while you don’t believe that only providing bandwidth will be able to capture value in the market, you would agree that telcos and mobile operators can only reliably derive profit from “the IP packet and IP addresses,” i.e. connectivity/bandwidth? Leaving the capture of value from software-based services to those, like Skype, who “crystallize into one thing 90% of what everyone wants?” 1. Despite Gordon’s enthusiasm for Skype: Forster: I guess it depends on how adept the telcos and mobile operators are. Probably some of them will make this shift. IBM was adept enough to shift quite a bit over the last decade. The mobile operators do have inherent value beyond bandwidth: location. They know about where the phone is and this can be valuable although it’s not clear if it’s a major value or how well they will capture it. It partly depends on things like the typical screen size on mobile phones. 103 Melissa Davis: To Richard Stastny I say right on point here in many ways. 1.1. No one in my communication groups or customer groups feels [any need] other than to configure their network desktop management systems to detect agents, destroy them, and report the offender for violating security policy, not against per se, but against using nonapproved applications, and exposing the corporate network to outsiders by maintaining easily exploitable open TCP port 80 connections. 1.2. Enterprises own their networks, the infrastructure, the machines, and the software that runs on them. Those machines are for business use, and the user signs an end-user agreement to that effect. This notwithstanding, the enterprise cannot dismiss liability for misuse, fraud, theft other than by due diligence based on industry standard, and judicially recognized as “reasonable” procedural defenses. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA 1.3. Enterprises have already gone down this path with Napster and Kazaa. Hence, they are not unprepared with policies, guidelines, practices, and corrective action. Use of Napster and Kazaa or other P-2-P tools are as punishable by dismissal, litigation to recover losses, and prosecution on par with importing or exporting sensitive information through email, ftp, portable mass storage device, or other more traditional means. 2. Regardless of whether enterprises could create a Skype of their own, I agree with both Richard and Malcolm that there seems to me no earthly reason or cost/benefit to so do. As Richard points out, commercial and government IT want, and in some cases are constrained, to buy COTS (commercial offthe-shelf) software. Advantage: products that are supported and can do SIP, H.323 transparently and use SS7/IP signaling when PSTN destinations or sources are required. Vendor maintenance and support are a necessity and a financial no-brainer relative to the costs of personnel and training. 3. Commercial and government networks are still accounted as a “cost of doing business”. That means, for those of us who have operated and run big networks, quiet and controlled. That is a fundamental difference from those who simply are “users” of networks, such as end-users. 4. My big customers have huge security concerns. Just one location with 7000 users reported 11 million penetration attempts in just one month. P-2-P applications, such as Skype and “GoToMyPC”, which, by means of their operations, bypass layer 3 and layer 4-7 firewalls and IDS/IPS sensors/detectors, allow the thievery, brigandery, highway robbery, vandalism right into the core infrastructure. 5. VoIP still is of primary interest, as a cost avoidance for replacing expensive legacy PBX’s, as well as other advantages. SIP, SIGTRAN (SCTP and SCCP-UA), H.323 are of interest. Those protocols fit with the ability to manage and control the network. 6. ENUM, as a directory service, to allow location of devices and persons associated by XML links to devices, call re-routing, call direction (e.g., next available customer service rep, or high priority service rep for platinum class users, or “if Jim is busy, ring Doug if free, else Sarah) still is a concept to be proven. 6.1. it must have authentication 6.2. it must have authorization attributes 6.3. it must be protected against the very kinds of vulnerabilities that fuel the daily escalating wars between vandals, thiefs, brigands and netadmins with DNS 6.4. it must be secure from RATs and tiny data miners that will make ENUM a greenfield for IP Phone Spam, identity theft, and “phishers of men.” some very preliminary discussions with a couple of large US system integrators. I am wondering if this is part of a larger trend? Davis: To my knowledge, there has been only a single instance of such to one of the Agencies of the U.S. Government. I know personally about that one. I have to tread very carefully here due to NDA’s and internal corporate constraints. UCLP in some form, whether directly Edge-to-Edge controlled or controlled by a rapid web interface OSS/BSS of a provider is imminent. The Canadian NRC, the US DOE/NNSA, DoD, Treasury, Justice, the US Homeland Defense are all constrained to maintain the most rigid security on their WAN links. Much of this is still shaking out. I don’t see a Skype vs. vendor based VoIP contention. Animals co-exist in the same eco-spaces if they are not competing for the same food source. For some people, and this includes me, we have opted out of the Microscheiss tax and security holes. That doesn’t mean that Microsoft feels threatened by us in its desktop market. I am an architect and an engineer. I leave the broad sweep vision thing to others, and avoid polemics. I think about and act on the next few steps. There being no money for me in a de-monetized application, Skype is just not something I think about. There is a major RFP that will be released in April for the entire Federal government. This one is called “Networx” (US GSA) and is a replacement for the older GSA “FTS”. Winning a selection place doesn’t immediately transfer to contracts, but it does free other Government Agencies from having to let bids. They can just buy from the list of preferred providers. Matson: Gordon - Many of us think that the major private enterprises simply haven’t begun to understand the implications of flat, universal global peer-to-peer communication has for their business models and very structural survival. I am not sure there is much to be gained by “blue-skyping” the future in this manner, but I have for years held to the words of the great Walter B. Wriston: St. Arnaud: Melissa: Agreed. Well said. Are you seeing any changing requirements for wide area networking because of security? Lately we have seen a couple of RFPs that will not accept MPLS VPNs or VPLS because of security concerns. This, of course, is music to our ears in terms of UCLP. IN fact one major project that we are deploying is linking all the NRC labs in Canada (Canadian equivalent of DoE labs - but on a smaller scale) with UCLP so that they can manage their own truly private network, and reduce number of firewalls etc. We are also in 104 “”The philosophy of the divine right of kings died hundreds of years ago, but not, it seems, the divine right of inherited markets. Some people still believe there’s a divine dispensation that their markets are theirs - and no one else’s now and for evermore. It is an old dream that dies hard, yet no businessman in a free society can control a market when the customers decide to go somewhere else. All the king’s horses and all the king’s men are helpless in the face of a better product. Our commercial history is filled with examples of companies that The COOK Report on Internet May June 2005 failed to change with a changing world and became tombstones in the corporate graveyard.” Maybe Wriston’s wisdom applies to a much higher level of the established order than just individual “companies” and “markets” in times of seismic technological change such as we are currently in. Davis: Malcolm, I totally agree with both of your statements above, but not the implication lying transparently in your syntactical construction, some indicative mood form of a French “n’est pas” implying that the audience should bow in affirmation. 1. I haven’t seen such business models. 2. Enterprises (private, public, government) have business models globally and very distinct component models, each with differing levels of need for the CIA triad (Confidentiality (sensitivity), Integrity, Availability). If your statement and your quote are simply techno-religious a priori value axiologies, of the now trite cliche of “Internet vs Telecom”, or “Power to the People and tear down the walls of the oligarchs”, or a moral imperative for the kind of transparency that makes long haul optical glass seem opaque: then we simply have nothing more to say to each other. I don’t share your axiologies, therefore discussion is impossible. The eco-spaces of end-users and their communications needs are fundamentally different from those of complex organizations, though for some businesses overlapping at the edges. Enterprises, whether privately held, publically held, or governmental have external constraints to protect and maintain confidentiality and integrity of certain information stores and transaction records. Consumers are not so constrained. Enterprises are playing in a competitive environment. Intellectual property, business strategy, and much of corporate governance must be maintained with restricted availability to “need-to-know.” The cost of loss of availability, as to propagating worms, is huge in any business, where, for the end user, loss may be inconvenient, but of minor financial impact. Per the following quote from the Skype web page: “We prefer to think of ourselves as a big group hug, even a present. Yes… that’s it… we’re a present… but without the ribbon.”: My customers and I retort: “Beware of geeks bearing gifts.” Skype, Firewalls and Security Coluccio: Also, what of my earlier implied observation above concerning Skype's‘ ability to penetrate WiFi hot spots’ firewalls, even when those hot spots are not part of such a promotion or ongoing service? Davis: Frank, perhaps I haven’t explained sufficiently how Skype, and other such “firewall-friendly” applications work (also reference “GoToMyPC”). As David Reed has said often here, the insider threat far, far outweighs the outsider threat. Skype is not penetrating the firewall from the outside in. Skype penetrates the firewall from the inside out, on the same TCP Port 80 or 443 connections that are open by design for web browsing. Skype goes out just as your Firefox or (hopefully not) Internet Explorer goes out to LightReading or CNN or Google. What is different about this breed of applications (and GoToMyPC) is that they maintain persistent strongly encrypted connections to external untrusted or untrustable servers. That may not matter to an individual end-user or a “hot-spot”. It does matter to any organization that is liable for at least certain categories of its data stores (by statute or to stockholders), privacy of personnel and medical records, employee behavior (e.g., harassment). It may seem “way over the top” for a corporate or government agency to have 105 to acquire, secure, and maintain TB’s of email stores, but the cost is far worse for not being able to answer the subpoena. Reed on March 8: Melissa - your point that connections to from employees’ machines to outside resources pose a risk is dead right. But the risk is not exacerbated by those connections being persistent or encrypted. The real problem is that internally companies are “wide open”, so any random person passing through (especially employees and their visitors) are exposed to information they shouldn’t be allowed to see. Making rules at the membrane surrounding the company is like creating the Maginot Line - you bet your entire security on the assumption that the path the attacker will take is where you think it is. Essentially this is because keeping the internal records secure *internally* is viewed as “too expensive”. This unwillingness to do the job right shows up in lots of ways. Why was that “backup tape” of Bank of America payroll records that made the front pages last week not encrypted when it was shipped on a truck and insured for a few hundred dollars? Surely the BofA risk managers know that exposing such personal information is important to stop. Similarly, you wouldn’t have to worry about people making phone calls through the firewall (even if the phone calls could be hijacked by VoIP hackers to carry other info), if the fundamental security principle called “the principle of least privilege” were applied consistently throughout organizations. Protecting an organization against losses should not be a matter of random stabs in the dark, worrying about symptoms and highly-hyped threats, but instead should involve keeping the valuable and highly private information inside an organization in a secure manner in the first COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA place. terminate. In other words, implement a security infrastructure *inside* the organization that limits access effectively. 5. In side the internal collision domains are internal SPF/FW’s and stealth mode IDS/IPS sensors. The main function here is to monitor the presence and epidemiology of worms, port scanners, unusual activity. The legal “procedural defense” rationale is to establish that user behavior is subject to monitoring. Now, you and I know that packet sniffing is a notoriously bad way to do something like that, given the sheer amount of traffic, but it is arguable and is argued. Then Skype or VoIP or whatever cannot be a threat in the first place. Davis: David, You are dead right about everything you have said, with a single exception that I, at this moment, disagree, but am willing to be further informed by your argument if you will advance it another few steps. Point I wish further clarification on immediately below: David Reed wrote: But the risk is not exacerbated by those connections being persistent or encrypted. But first Per agreements: 1. Simple “Maginot Lines” of perimeter security are of little usefulness, for enterprises with the beaucoup bucks they spend on SPF/FW devices. 2. Inside the “RED ZONE” (Public Internet) perimeter are the email and web packet content inspection and filtering engines, the publicly facing Directory Servers (Border Directories), SIP Proxies (if exist), Web Proxies, and IDS/IPS (Intrusion Detection Systems/Intrusion Prevention Systems), TACACS+ and RADIUS RAS (Remote Access) services. The Citrix Nfuse and ICA servers would sit here, if exist. For medium companies, this is a single DMZ. 3. For larger enterprises, the description in #2 above, exists in an Orange Zone DMZ, handling traffic primarily facing the Red Zone, or the Public Internet. The main ingress and egress MTA’s would sit here. 4. Behind the Orange Zone is a Yellow Zone DMZ, primarily handling traffic and transactions within and between the various Green Zones (considered Trusted Zones). Intranet Web Servers serving as secure portals for internal transactions (e.g., finance, HR, labor, executive strategy) would be here, Yellow instead of Green if the Enterprise wishes remote access. Or they may be in Yellow because that is where the inter-data center links 6. The IDS/IPS systems are also there, along with the NMS (Network Management Systems) to map traffic in effort to discover “rogue gateways” (illegal or unauthorized gateways to the Public Internet). 7. Internal NMS monitor switch and especially router interfaces and buffers, again to detect the kinds of congestion or malformed packets that would indicate a potential packet pandemic condition. 8. Internal workstations and mobile laptops have virus scanners on them. With internal workstations, the scan engines and signature updates are pushed out with software transparently to the user. 9. Laptops are a huge problem and are getting more and more attention as a political battle focuses on locking them down vs “creativity and innovation.” Increasingly, mechanisms are being planned and experiments being done with requiring scanning before the login is consummated. That is, with a login request, the user request is placed in a queue while another process patches and scans “out-of-compliance” machines. 10. All of this, as you know, should follow a thorough assessment of what infrastructure and data resources an enterprise has, triaged by risk (impact of loss or compromise). It is here, David, that your major point of “least privilege” applies. Is it done? In my experience, yes when it comes to infrastructure (routers, switches, server sysadmin). 106 My experience with application access, and most particularly data store access, is that it is much easier for a business unit steward to grant privileges than to revoke them. If there is a formal employment adverse action, the revocation is not so difficult. Absent that, and particularly in “white collar union” shops and civil service, one could be challenged to meet almost criminal proof standards to sustain such a revocation. Further, particularly North American enterprises have largely followed by a long way that of major European enterprises in adopting X.500-like LDAP Directory Services, which allow authentication to be done globally with authorization done locally (with a few mouse clicks). Consistency checks and date checks can be run to produce alerts on possibly stale permission attributes. David Reed wrote: But the risk is not exacerbated by those connections being persistent or encrypted. Davis: David, here is where I may be blinded by my own training and the cybersecurity culture in which I immersed myself after the Bust. 1. the persistent, always on, connection seems to me vulnerable to any number of hijacking, session key theft, and intrusion techniques through known exploitable applications, e.g., Windows Media Player. 2. the persistent connection is simply out there hanging for any black hat to use as a route of propagation for any kind of malware they have, again, Windows Media Player is an excellent example of a propagation route 3. if I allow a non-encrypted connection, even locally to a proxy, I have the snooping ability to investigate suspicious activity (an insider exporting protected information or a “captured” insider machine with RATs, Zombies, Trojans, Data-Miners and other digital critters. 4. perhaps not in general with , but definitely not with encrypted sessions, can I enforce what I am compelled to enforce, specifically calls which are sexually ha- The COOK Report on Internet May June 2005 rassing, conspiratorial, etc. Please tell me what I am missing. Reed: Melissa - Disclosure doesn’t require persistent connection. A novel’s worth of data (plenty if you are trying to capture a corporate strategy) is only half a megabyte or so. Given the number of bits flying in and out the doors of any company, that takes an instant. It fits in a tiny corner of an otherwise innocuous document, perhaps by misspelling occasional words in emails over time. It takes a few *milliseconds*, not hours. Disclosure doesn’t require obvious encryption. At the network level, you or anyone else cannot understand what 99% of the bits mean, or understand what bits are reasonable and what ones are not. It may be that you think you are mandated to listen to everyone’s phone calls because “sexual harassment law” requires you to do so. I understand that that’s the interpretation of many lawyers. Independent of my judgment of whether that’s good public policy, it’s also technically stupid. You cannot assign enough people to the job to notice and block all attempts at sexual harassment merely by listening to phone calls. Even if you hire a whole staff of outsourced Indian English-speaking listeners at low dollars per hour! The same argument applies to the idea that you can monitor every possible word flying through the network and prevent theft of information. The answer to blocking theft of information is to control the flow of information to the small number of people who need to know in the first place, not buy thousands of dollars of equipment because you want to buy Microsoft’s garbage systems and patch them around the edges because they do not allow you to build appropriate security policies at the right grain in the first place. Davis: We have no disagreement here. Your words below capture the problem eloquently. Reed: The answer to blocking theft of information is to control the flow of information to the small number of people who need to know in the first place, not buy thousands of dollars of equipment because you want to buy Microsoft’s garbage systems and patch them around the edges because they do not allow you to build appropriate security policies at the right grain in the first place. Davis: As for the monitoring, again we have no disagreement. As a practical matter, no team of people is large enough to monitor the volume of transmissions and content filters are easily fooled with syntax (e.g., double negatives) mis-spellings, and shared secret word substitution. In the case of suspicious behavior, with internal crypto, such as PKI, the stream could be encrypted and the transaction time recovered under legal or other policy permission. But for now, the main points are those of the lawyers and the “procedural defense”, as well as the fact that these enterprises continue to pour big dollars into defective insecure operating systems. Skype in the “Civilian World” Schulzrinne: “Civilians” do get some of the power and capabilities that larger organizations had a while ago. This tends to broaden availability, but there is little evidence that the same technology implementation is appropriate for professional use or somehow puts the professionals out of business. They definitely tend to exert downward pressure on prices, as 3Com found out when companies like Dlink, Netgear and Linksys entered the LAN space. Cisco and Extreme Networks seem to be doing fine, however, and I haven’t seen a large organization throwing out their Cisco gear for Linksys switches. (I know that Cisco now owns Linksys.) The demise of telephone service as a major revenue generator for the ILECs will likely have little to do with softphone clients like Skype and much more 107 with cable companies offering this as an add-on service, in addition to Vonage, Callvantage and the like. The interesting questions, financially, is not how many people use softphones like Skype, but how many would be able and willing to forego paying the phone company or its replacement. It is indeed likely that the total amount of money spent in corporations on PBX, IP or traditional, is likely to decrease longterm, but that hasn’t been a major factor in corporate IT spending for a while. (We have had the same classical PBX for twenty years, so the likely outcome is that we’ll spend less money on the next generation, but more often, as I can’t imagine that we’ll use the same IP phone and software for the next 20 years.) The notion that peer-to-peer somehow makes Skype fundamentally different or better is naive, at best. The cost of supporting a SIP server, for example, in an enterprise is trivial, compared to end systems of any stripe. (We have two redundant Sun low-end servers, at about $1000 a pop, for about 200+ users. They could handle a large part of the university if they had to.) Davis: I want to look at ENUM and Reed’s Law and then build on Henning’s economics: Henning presents excellent points in his paragraphs quoted immediately below, especially with regard to the economic case. The collapse of long distance pricing has long since eliminated the toll-bypass economic arguments for VoIP of any kind. The collapse of long bandwidth and peering prices have reduced the cost of operating a voice network on the legacy circuit switched paradigm to near cost or below. Flat rates for access and flat rates for number of minutes pervade the circuit switched business pricing policies. Legacy PBX’s, per Henning’s observations, largely function if supported by maintenance contracts, are rarely a security concern, and provide a comfort- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA ing and economically desirable transition time for the businesses owning them. It is the rare commercial or government enterprise that will be stuck with the label “early adopter.” The cost of being wrong relative to the probability of being right are just too great for careers. Now to ENUM and Reed’s Law as it affects Skype and the predictions of scaling in customer uptake: Reed’s Law arose from a refutation of Metcalfe’s Law (the n-wise possible connections being the familiar {N*(N-1)}, where, as N scales to “large”, approximates N^2 as a limit. Reed, without giving the name of the female engineer elucidating the refutation, paraphrases her as showing that if N people make one call per day, the communications pair-wise connections approach N, not {N*(N-1)}. Thus, Reed reports her conclusion, inevitable by her analysis and his reflection, that it is the size of the network, not the pair-wise connections on the network at any given time, which produces revenue. Reed knew that of any group of N members, 2^N possible groupings can be formed. In group forming networks, Reed concludes, the connections that a network can enable then is 2^N, where as N grows, quickly becomes to dwarf Metcalf’s {N*(n-1)}. MELISSA’s LIMIT ON REED’s LAW (and by corollary a design requirement for ENUM): Both Reed’s Law and Metcalf’s Law are simply simple two dimensional infinite sequence models, necessary from the algebraic rules on which they are formed, but in no sense empirical, as laws of physics, laws of cellular automata of > two dimensions, or of complex systems, that is dissipative systems. While it is easily shown mathematically, to the credit of both Reed and Metcalfe, that it is possible to grow some infinite network to infinities, it is equally clear that neither Reed nor Metcalfe factored in any dampening variables – in other words boundaries with gradients. Communications networks are connections of humans, thus are as much a psycho-social and psycho-physical phenomena as a technical ones, and the former pair must exist with interest before the latter. This is to say, there must be players in the game, playing under conditions (rules, boundary gradients) which return a value to them for the dollars and time they spend. If ENUM is to work, one must be able to control one’s own entries, that is, to turn it off for one’s own protection and peace of mind. In the PSTN, we have the option for private, unlisted numbers, call blocking, etc. By now it is well known that the O’Dell curve on Internet growth that fueled the speculative bubble was a statistical fluke, that is the growth doubled every ninety days for the two quarters under study, and as far as is known, did not before or since. Some of us do not want that, celebrities and otherwise. We have so much communication that we have difficulty being productive, not the other way around. As a psycho-physical model, Reed’s Law doesn’t map to what we know well and incontrovertibly about humans and our great ape cousins. We not only are not “on” for psycho-social contact (including cyber-psycho-social contact) all the time, but past a fairly stable mean, increases in psycho-social “connections” become aversive at an accelerating rate. This would turn both the Metcalfe N^2 and Reed 2^N exponential curve into a downward opening parabola. Too little “connection” (pair-wise or group), and there is a rapid uptake to grab for more. Beyond the peak and duration window (we have to map with Fouriers), and interest declines with annoyance and irritability building. This suggests to me that, Skype and other “always on” kinds of vulnerabilities to “over-communications” will set a limit on the scalability of the medium and its rate of adoption, absent any other dampeners. The other dampeners are risk of compromise to one’s own communications device, already well known as a highly preferred epidemiological route of propagation of spamming email relays, email address capture for spoofing, RATs, dataminers, worms, zombies, and other ugly digital beasties. ENUM: 108 Malcolm stated earlier this morning, and other Skype fans tout, the “flat, peer-topeer, global communications.” But Skype May Show up in Some Enterprises Sooner Rather Than Later On February 20 Coluccio: Over the past two weeks, on an unsolicited basis I’ve become aware of two instances, merely by speaking with acquaintances, where Skype is actively being piloted by a very large business enterprise. In one case, I was told of a publicly traded semiconductor manufacturer that has discovered some flow-related issues that have arisen inside its firewalls, which is looking at Skype and should have a fix on, soon. In another, a financial institution on the East Coast is readying a launch of an application similar to Skpye (which was hinted might even be Skype dressed up for the occasion as something else), but no word as to when that might be. Shockey: I’ve heard this as well. Skype is spreading like wild fire in European circles. James Enck the noted telecom analyst with Diawa Securities has noted this extensively in his blog http://eurotelcoblog.blogspot.com/ Coluccio: And clearly, Skpye has had a very active front office, despite images one might conjure by statements in the press about its limited size and presence, as witnessed by the number of very serious announcements that have come out over the past week, alone: a broadsweeping announcement with Motorola; Amazon and Wal-Mart will be hawking The COOK Report on Internet May June 2005 it i.c.w. a PC bundling arrangement with a Unix OS and Firefox; and discussions surrounding a host of mobile devices that would become Skpye ready, soon, if not already. were originally intended, for some time to come, and within designated zones of trust. So much for the blue chips’ and the government’s concerns over security and reliability. IM was once regarded with the same level of skepticism and mistrust in large enterprises as Skype receives today, but over time it has gained acceptance, even at the cost of losing its identity during screen pops, yielding to the monikers of the firms who employ it. And in those cases where IM is used, in some cases by the nation’s largest banks and brokerage firms, they are indeed designed to operate as walled garden applications, barred from receiving or transmitting anything to or from the outside world. Shockey: See the Microsoft LCS platform. What you described is being delivered. Shockey: Excellent Frank. I recently wrote a magazine piece pointing this out. Skype, Web Services and Mission Criticality Coluccio: Having said that, during the near to intermediate terms, I cannot see either IM or Skpye being used to fill the shoes of mission critical applications or any of the more advanced requirements that are now finding their way into complex Web Services. Shockey: They will once they become standards-based. Web Services have completely taken over the data exchange market because everything around web services is based on real documented standards. Coluccio: Instead, they will continue to be used as a form of expedient to save time and steps to get the job done, by taking the place of email and phone calls. And this will be so NOT because they cannot be made to perform to fulfill the more advanced functions, but for all of the reasons that Melissa has stated. But I do believe that they will find acceptance when they are modified and/or tuned to the organization’s specific criteria as related to both security and their boundaries on reach, and will receive use only in the manner in which they Coluccio: For the other ninety some odd percent of the universe, however, they will continue to adopt Skype in the same ways and for the same reasons they’ve ever adopted anything else. If it works to a user’s favor, it will be adopted. If not, then it won’t. There are tens of millions of small to medium sized businesses out there whose sole connections to the Internet consist of one or two DSL lines or a T1 line or two. Add some tens of millions of consumers who attach via broadband from home, and now from the street via certain forms of wireless, as well. And then we are hundreds of millions of citizens of all types who overlap with at least one of the categories of presence listed above, which means that you can then mark some of those individuals down twice, sometime three times, as candidates for using IM, or Skpye, or whatever the next pop up act is that comes along. Several days ago I had an interesting experience. I started out by having a long business conversation over Skpye, which migrated to my cell phone when my laptop audio bummed out on me, as it sometimes does without notice, and then to my cordless POTS phone when my cell phone’s battery showed signs of expiring. During that one and a half hours, approximately, I used the last mile services of at least three service providers and the Internet, without counting any of the services on the other end of the call. And consider, I am both a private consumer and the owner of the consulting firm whose business was being discussed during that call, and the party on the other end of the line(s) was a client of mine (who insisted on using Skype in the first 109 place!, and hasn’t stopped thanking for my introducing him to it, yet). How many times do I get marked down for my participation during that single, albeit multi-parted, conversation? Hassinger: Gordon asked me to weigh in on the web services aspect to VoIP and Skype. I see web services as a way to loosely link application components - think distributed computing with less of the painful complexity that guaranteed it would never be taken up en masse. In that regard, there is clearly a whole suite of voice-centric componentry that will be required for the types of composite applications that enterprises (and end users) will be wanting to instantiate. These components and their boundaries are not yet known and won’t be until we get into full swing of implementing composite, web services-based applications. They may range from versatile client front-end to voice communications to backend ‘message center’ workhorse to generic ‘speech-to-text’ modules,’ all for mixing and matching into cross-enterprise, category-blending applications. It seems likely to me that the components will need to be compliant with the dominant standards for interface, provisioning, and management (at the moment these are Web Services, however I think we still have some turns of the evolutionary crank to go before we get to the ‘final’ protocols for composite apps). Like the Web itself, however, the approach is very much retro-fitting, and therefore it is likely that a technology/ component will emerge as ‘best of breed’ *before* it becomes a standardized plugand-play part of the Web Services infrastructure. Why? Because loose coupling and late binding don’t amount to much of an incumbent advantage. If there were a Web Services component for SIP calling in use today and Skype became indispensable tomorrow, you could just slap a Web Services layer onto the Skype client and swap out the existing component. The long and the short of it is that the future of enterprise computing is look- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA ing more and more like the capricious, faddish consumer landscape. If Skype or whoever convinces the world through viral marketing that their technology is a ‘must have,’ CIOs and IT departments will have pretty feeble arguments to opposing that urge. The very fact that we are discussing the merits of Skype, a p2p app with a lineage that traces directly back to Napster (through Kazaa and gnutella) proves this point, at least to me. (It goes without saying that these are my personal views and do not reflect those of my employer.) Reed: Sebastian’s point about late binding and loose integration is important. Some control-oriented suppliers (the usual near monopolies) and control-oriented IT dept. customers, see “Web Services” as the new SAP - the procrustean bed that all corporate services have to be forced into. But pragmatism suggests otherwise - the key thing about successful companies is that they retain flexibility to learn new things and do new things, and the sensible use of Web Services technologies can achieve that. Only the most bureaucratic companies let their IT departments dictate how the lifeblood of their companies, information, should flow in their arteries. IT depts. that focus on creating business opportunity (such as new ways to connect to customers and suppliers and partners) are the ones that help their companies succeed. IT depts. that resist new ideas in order to maintain control often use “security” as a codeword for anti-strategic action. It’s a paradox that is very simply endto-end secure, and end-to-end resilient compared to such networks as the POTS, cellular, and SIP+CALEA networks that are compromised from the start, but corporate America is making up stories to try to paint it as insecure and unreliable. Skype has a problem, in that it is not currently able to openly demonstrate its secure nature (potential, and probably actual, because reverse engineers would discover and expose the holes rather quickly as they have with SIP implementations), because its protocols are proprietary and somewhat secret. We had exactly the same problem with Lotus Notes when I was at Lotus, by the way. When the US government required Lotus to weaken Lotus Notes crypto for export (after I left) the proprietary code in Notes made it vulnerable to accusations that the US was using it to spy on non-US companies, and for all I know, it was doing exactly that (my more grandiose spook friends hint that US suppliers have private deals with the NSA to put backdoors into every product that is marketed overseas; I don’t know if I believe them, having never been asked to do so at Lotus when I was in charge of such things - I didn’t verify that some junior programmer didn’t do exactly that). However, that point is true in spades for the SIP suppliers, because its security is still dependent on every component of a solution working predictably and correctly. SIP has no “brand” investment in security, and it has no enforcement over its implementers. Sip Based Enum Wi-fi Phones On February 21, Dewayne Hendricks: I just completed an iChat with James Seng, who’s attending the APRICOT 2005 <http://www.2005.apricot.net/> conference in Kyoto. James pointed me to his recent blog entry titled ‘Giving out WiFi SIP Phones’ <http://james.seng.cc/ archives/2005/02/21/giving_out_wifi_ sip_phones.html>. If you’re one of the lucky folks attending the conference, then you would have received one of the new Hitachi Wi-Fi SIP phones as part of the ‘APEET ENUM/SIP Live Trial’ <http://www.apenum.org/APRICOT2005/Live_Trial>. Take a look at his blog entry and consider the future as more of these phones start to roll out with support for things like ENUM. I have had a Pulver Innovations ‘WiSIP’ phone for several months now, programmed to use my Vonage softphone account. I haven’t made much use of it on trips due to the problems with getting it to register with commercial hotspots which require web access of some sort to authenticate and use their service. Hopefully, these next generation phones will make travel use easier then it has been to date. 110 COOK Report: Do read the blog http://james.seng.cc/arentry chives/2005/02/21/giving_out_wifi_sip_ phones.html . Davis: SIP Enum is a viable option, through a gateway, mediated by a circuit layer proxy, to an IP PBX that allows internal users to select availability options and what will be published . . . and to provide this with authentication and individual+role/group based authentication and authorization of voice mail boxes via directory services, is a buildable service via metadata/XML. We haven’t seen such yet. But the consumer end-user market is a different eco-space. There may be differential communications/IP within even a small business: per our phone conversation, use of a Nextel VoFR walkie-talkie may be all that is needed in calling from the retail section of an Ace Hardware store to the storage area in the rear. A different kind of communication and security may be selected for credit card point-of-sale transactions and EDI interfaces with suppliers. In an interconnected world, there can be connected communities of interest with walled garden protocols, but the wider the community on some scaling factor and the more complex the work through applications, and the changing variances on the need for confidentiality and integrity, the greater the commonality of protocols . . . . . . yet, the greater the commonality, so also the greater the risk to any exploitable vulnerability. We just don’t know how it works. SIP servers are a risk, but one security officers more and more sign acceptance of such risks --- because there is mitigation that can be taken. At least this link didn’t advertise itself as “fire-wall friendly”. ;-) If you don’t know what this means, it means that the traffic of that application (e.g., and GoToMyPC) slips through firewalls with outgoing http, and the end-to-end encryption makes the information flowing impenetrable to IDS or Firewalls. The COOK Report on Internet May June 2005 Hence, in a crafting of words as skilled as Carl Rove’s mining of Red State prejudices, “firewall friendly” is from the reference point of the application, not the reference point of the people who are charged with defending the integrity of the network. Reed: If corporate IT thinks that security is enhanced by making the internal VoIP system non-interoperable with the “public” voice systems, that is not surprising. However, it’s why corporate IT shouldn’t be allowed anywhere near business operations. The reason they call trade “intercourse” is because it involves intimacy with your customers and your suppliers. A business that can only talk to itself will die the death of the onanist. Davis: David, I don’t think there is any intention for that. Onanism refers to one person and one only. Any business has communications needs that are layered. Any business constrained by the need to mitigate its risks will be observant of those needs as they are now. Corporate competitive strategy and the latest and greatest product prototype secrets are not put in the company newsletter, nor are executive strategy meetings broadcast on Video/IP. The substitution of an application or a technology for the work to be done with that application or technology, and the scope and boundaries of that work is to substitute the means of transportation for the arrival at the destination and coordination of what is necessary to get there. Coluccio: David, I understand your point. However, I submit that our friends in the black hat cadre have another view of the term “intercourse.” And they appear able to “do it” with impunity, wherever they see an open orifice. On Walled Gardens and Getting to the Other Side of Geoffrey Mooreʼs Chasm Something Melissa stated earlier caused me to reflect on my earliest use of a closed, purpose-built network, which in some ways resembles the kinds of walled garden networks we see in enterprises today. During the late Sixties I found myself using a network set up via submarine cable channels and satellite links meshing NY City, Paris, London, and White Plains. Its purpose, like so many other industry ‘order wires,’ was to facilitate the internal workings of telecom company employees via express means, sometimes ‘nailed up’ to overhead loud speakers in work locations, similar to - and in some ways identical to - hoot and holler networks used by the junkyard and trading floor crowds. In the case cited above, every channel that was used was a dedicated full-time circuit independent of the TASI (time assignment speech interpolation, which was the analog precursor to digital speech interpolation techniques used in early TDM T-1 multiplexers) based network it was designed to protect, thus performing the function of an independent overlay that could be used during times of cable, satellite and TASI system failures. While these are still used extensively today (see url below), one might think of a wireless based public safety and service network in the same context, independent of public service provider facilities. In my earliest contemplations of how Skype would be used in a commercial context order wires were the first things to come to mind. From: http://www.dsptele.com/products/ order_wire.html “An order wire is a voice communications system used primarily by maintenance personnel to communicate between equipment sites. It is a telephone system, but has no central office switching. It uses a portion of the network’s bandwidth that is not normally used for revenue traffic, such as the bottom portion of the base-band spectrum or a portion of an overhead bit stream. It is an economical phone system that is network owned.” 111 With substitution of a few words it’s easy to see how my view of an IM-like voice application might be supported by the definition above. Wetzel: Just my humorous point of view regarding what I think of the corporate sector. I believe nothing good, new, or technologically innovative will ever come from that sector. They always will be followers and join the wagon very late when success is here. IT people in that sector are worried about security issues only because they want to keep their job or protect themselves from their N+I bosses. They sell them a firewall that lets everything pass and they will be happy because they got their firewall. Davis: Damian, Geoffrey Moore, in his latest and very popular two business books on techno-business articulates the same. But, one wonders why you think it should be different with business and technology, vs smart money in general. My grandfather taught me well that “smart money is conservative money.” Moore’s analysis of techno-ventures show even most successful ones falling into his now famously named “chasm.” Most of the other successful ones are acquired by big techno-corps that have the incubators to hone and refine, to focus horizontally on the vertical success. Money and the bulk of the purchasing power is in the hands of the “middle adopters”, the “middle late adopters”, and the “late adopters.” Early adoption, where large consequences for failure or large costs on impact of operations and labor are at stake, is simply foolish from a cost/benefit point of view, and dollars rule. Damien, I am not saying I have never been “an early adopter.” Hardly, but what I am saying is that I am “an early adopter who was mugged.” Architecture doesn’t upset me. Skype is an entity. Worms/Trojans/RATs/DataMiners/tiny NMAP port scanners are entities. GoToMyPC is an entity. Microsoft is an entity. AT&T will soon no longer be an entity. The old IBM and HP are scarcely visible in the present incar- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA nations. Cisco is no longer getting 90% margins on cheap Motorola and TI chips and NVRAM. [Skype] will get traction or it won’t, in some niche, some niches, and over some duration. That isn’t an engineering decision or an architectural one. When it begins to interfere with business revenues, it will be come a judicial, regulatory, and legislative issue. I don’t know how Skype will turn out. ‘Skype’s success would be a huge push in “the race to zero.” Depressions, deflation, stagflation are all about perceptions of economic and institutional instability. Capitalism works because the big guys stay big, there is enough room in this vast economy for experimental innovation, and in times of economic and institutional stability, there are incubators for those experiments, and time to allow them to get traction and modify themselves to fit. Revolutions impoverish everyone, and usually for a very long time - as Japan, Inc has proven. The shock that has rendered the Japanese stagflation impenetrable to monetary and industrial policy is the shock to the culture of the collapse of the real estate and banking enterprises, which were backed by a government/MITI that had so long promulgated invulnerability and wisdom. You are seeing now in the telecom field what has been predicted. You are seeing what your anarchistic leanings fear the most . . . consolidation, or re-consolidation, “pay for play.” IP won as the transport of necessity now. No one cares, especially the ILECs about the circuit switched legacy dinosaur and all are moving to max revenue generating businesses and services away from “common carrier.” FTTH is dead except as a symbol - making almost no economic sense with a single exception - the ILECs see their competition for the SOHO and residential customer as the cable co’s, and each other as competitors for the large enterprise/government space. Services, the ability to deliver a branded product with a stable provider at a reasonable price, is the game for SOHO and the bulk of the non-early adopter consumer. You forget how techno illiterate the consumer is. In the enterprise/government, bandwidth is simply assumed to the edge. The differentiators are about service delivery and SLA’s, as well as MTSP (MultiTier Security Planning) deployments that cut the internal costs to the enterprise and fovernment customer. I don’t need a solution, Gordon. That isn’t my game, predicting the broadsweeps of some future. I am an engineer and an evolutionist. Evolution doesn’t reward more than very few of tens of millions of mutations, and those are small ones, mostly neutral in terms of survivability and incremental improvements I don’t know what optimization even means. I know that what I have is a niche is grey matter, experience, knowledge, and the self-discipline to get more and make that activity fun. So I only think about getting through the next few steps and how to get a decent check for that. Being a player in the “race to zero” is not in my interest, personally or professionally (by association). Skype and Grid Computing Having been an early experimenter, and participant in research, with Grid Computing, I had been thinking of possible parallels of how Skype or Skype-like unified messaging might occur. In the early days of Grid Computing, what we saw was an almost purely horizontal play with compute clusters, the grid controller’s only function being to manage connections, send and receive batches, order packets, and transmit when all batches complete. Grid Computing is evolving into vertically integrated work, off loading compute-intensive work from the applications with which users interface, sending results up to those applications (e.g., huge financial systems, nuclear emulation systems, huge financial-engineering applications integrating project manage112 ment schedule, budget, timeline dependencies with costs over several to many scenarios). The discussions here on Skype led me to such thoughts, especially after Jim Forster of Cisco suggested I try it. Per what Frank Coluccio and Jim Forster have written, there is the possibility to modify or reverse engineer the code such to eliminate the persistent “(ugh) firewall friendly” problems. We have done so with internal-only IM with near wire speed Web packet filtering/ACLs (Access Control Lists), VLAN design, and routing policies. So, it was with that sense of serendipitous propinquity that I chased last night’s link in slashdot to the following interview with David Worthington in Beta News on the evolution of Grid Computing. I share it as a possible path or at least vector of Skype evolution. Interview: The Future in Grid Computing By David Worthington, BetaNews February 21, 2005, 11:41 AM http://www.betanews.com/article/Interview_The_Future_in_Grid_Computing/1109004118 INTERVIEW Computing grids are software engines that pool together and manage resources from isolated systems to form a new type of low-cost supercomputer. In spite of their usefulness, grids remained the plaything of researchers for many years. But now, in 2005, grids have finally come of age and are becoming increasingly commercialized. Sun Microsystems recently unveiled a new grid computing offering that promises to make purchasing computer time over a network as easy as buying electricity and water. Even Microsoft is said to be investing in grids and Sony has grid-enabled its PlayStation 3 for movielike graphics. As interest in these distributed technologies grow, so does the probability for disinformation. With that in mind, BetaNews sat down with some of the The COOK Report on Internet May June 2005 world’s leading grid guru’s, Dr. Ian Foster and Steve Tuecke, to set the record straight and divorce grid hype from grid reality. BetaNews: Since we last spoke in 2001, what significant developments have there been in the commercialization of grid technologies? Dr. Ian Foster: Back then we were just seeing earlier interest in grid technologies from companies like IBM etc. Since then we have seen tremendous growth and enthusiasm. And a lot of things are being labeled as grid that perhaps one could argue they are not. Perhaps they are more, in some cases, computing cluster management solutions, but also some substantial early deployments in the industry from companies like IBM and Sun, and others like HP and so forth. [snip] COOK Report: Thank you, Melissa. At the end of my conversation with Stuart we talked about Skype’s possible evolution. Stuart talked about the advantages of a federated system - which system would incorporate the principals of Goroshevsky’s design for Peerio - the design could be done regardless of Goroshevsky’s lack of credibility. Federations are mentioned in the grid discussion below. Certainly anyone can reverse engineer Skype and probably should do so - including perhaps Skype in its next release giving itself federated capabilities? Part of the attractiveness of Skype is the tool set it comes with. Skype really does change the nature of one’s communication patterns - I am glad to hear that you will be giving it a spin. Last night both Frank and I were looking at the web doing slightly different google searches in real time and swaping urls back and forth via IM. Something not doable with the same effectiveness via email. While the architecture of the current Skype carries risks - undoubtedly- what can be done with it as a platform independent softphone in consumer hands way exceeds what the other voip models, tied as they are to the handset and the legacy telephone system, can do. I think this bears close watching. It will migrate to businesses for sure. How far how fast remains to be seen. The Google Telephone Network and a World of Abundance in Communication And back to the Google dark fiber discussion (pages 71-76 above). Have you all looked at Google’s new local feature and ‘s voice mail? Look at the maps available now on Google. It is much better than map quest. You can query those maps for all hotels within five miles of any address. Why not gas stations, restaurants, in short why not any kind of specific retail business? And why not the next feature being a Skype or Sype like calling number for that retail business? There is a potential here for removing vast amounts of phone traffic from the PSTN to the Google telephone network. I am just try to see some of the possibilities of where this could go. But the cautionary pushback is a welcome signal to choose my words with care. You have mentioned a race to zero that I find frightening indeed - but it seems that something - like it or not is happening - and the question becomes what are the activities in the society that add value and people will pay for? Or if not this what does the question become? Davis: Gordon, here is what I am looking for: simply a business model that would provide sustainability, scalability, direction, control. In all of anthropological research, there are only a few sustainable models that work. There is the techno primitive fairly flat, family and extended kinship of the subsistence hunter-gatherer, the warlord, the feudal cum nationstate model. There are equally only a few economic models that have shown workability, mass socialism of a Voltairistic model and mass Libertarian models falling to state or economic oligarchy (and apparently in the Western World, so do republics, parliamentary and the US 113 model). Open Source, e.g., Linux, is not the freevolunteer cyber peer-group of yore. Open Source programmers, and “the maintainers” of significance in the community now are largely employed by Intel, AMD, HP, IBM. Linus and McVoy are corporate animals of this consortium. So, my question has to do with the economic model. Given that significant economies are all monetized, what is the financial incentive? How far it penetrates into the realms of the techno-barely functional and technoilliterate remains to be seen. How it morphs to provide a border gateway for enterprise internal security is another open question. Per David Reed’s challenge about business connecting with customers, the incentive is to have customer and sales call centers totally available by any of the individual and hybrid media present and emerging. From the network perspective, those can be air-gapped, or DMZ segregated, from the highly sensitive inner realms of the enterprise? The Skype security issues for transit traffic through an enterprise boundary are not unfixable. Declared port ranges connected to SSLv3/TLS proxies would eliminate the persistent open connection from the enterprise internal machines to the “open external relays.” The enterprise could operate a filtered relay at the edge. Web packet filters could handle the transit relays with ACL’s. There are a lot of ways to make these two eco-spaces compatible. In my opinion, “Either/Or” zero-sum games are not going to be productive. Sterling: So just out of curiosity what is the Skype business model? Agenda? Skype endgame? Skype exit strategy? Reed: Well, given that Motorola just announced a Skype cellphone, and Skype Out makes money, I think the business model appears to be quite simple. Give customers something they appreciate for free, and they’ll buy other things from you, and people will value your brand- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Of course such a business model may not appeal to those who think business is about coercing customers and screwing them out of their last dime once you’ve bought the right to exploit them from the government franchised communications commission (funny how those initials crop up...).... think the views of the backers may be somewhat different, and it’s important to note that the initial backers, Draper et al, also flipped Hotmail at a relatively early stage. My guess would be a trade sale to a major Internet player, and my intuition tells me Google, though they’ve never done anything remotely as big as this on the acquisition front. I think an IPO is unlikely in the current climate. But many other businesses work exactly that way, just not the “business” satirized as TPC in The President’s Analyst. :-) Henshall: This came across my desk today. I know people that have been trying to do aspects of this for years. Matson: Well said David. You have it absolutely. Knowing some of the London Skype team, it is clear that they deeply believe that if you have a relationship with 75,000,000 people as a result of giving them something of use (at no charge), then using their creative imagination to develop new and unimagined elements of value for sale for a few cents/$ is not too difficult. They have really begun to understand the world of ‘abundance’ and the new business models which this creates - rather than the world of artifical-scarcity created and so vehemently maintained by the twighlight industries of telecoms, Hollywood, and music recording. http://www.twilightutilities.com/ and http://www.twilightutilities.com/Forwarder.html name. James Enck: Absolutely spot on. Zennstrom has said that if Skype could generate the kind of relatively paltry ARPUs that Yahoo! generates from incremental services (I think the figure he quoted me was something like $8 per annum) from a small proportion of users, it would be immensely profitable and they would be satisfied with this. My other observation is that Skype insiders have expressed to me their surprise at the rate of take-up of Skype Out, i.e., revenues are ahead of budget. My conversations with Zennstrom have always led me to believe that his personal goal is to be recognized as a visionary, and someone who helped to break down cozy cartels. He started out at Tele2 in Sweden, where this mentality is deeply ingrained. He got close with KaZaA, but it all ended in tears, litigation, and a hasty exit. So I think from an ego perspective, he would like to run it and see it get bigger, along with his profile. However, I “Skype to phone and phone to Skype. Never miss a Skype call again. Skype Forwarder acts as both a telephone and a Skype answering machine, as well as a Skype to phone and phone to Skype gateway. Incoming phone calls can route to TAM (telephone answering machine) or a Skype buddy, while incoming R calls can route to TAM or phone. Have a Skype call dial your cell phone, or dial a Skype buddy with your cell phone.” I’ve not tested this. Seems they want to enable everyone to have their own personal software PBX. The day is just getting closer. I’ll get a test or two done. I’m “SURE” this is not “enterprise” ready! What I am observing is “Skype” is creating interest in audio that goes way beyond just phone communications. As software engineers tinker with what can be done via windows and how to bypass its crippled audio systems and so on, lots of creative new applications are beginning to emerge. Skypeʼs Impact on Voice Traffic and Open Source VoIP PBXs O’Leary: Given the observation (above) that Skype is driving creation of new audio applications, how much of Skype’s traffic generation is new applications and how much is replacing traditional long distance? I expect this is pretty hard to sort out, but it appears that the Skype 114 folks provide stats regularly for their usage, as does Vonage (and there are VoIP market share stats for Vonage). Is “traditional” LD traffic still growing? I recall seeing various stats in the past the voice traffic was growing at 2-3% per year in the 1990s, but I believe this was all voice traffic minutes, not LD specifically. What does the LD growth rate look like overall now, given the continued erosion of prices, and how much of the overall growth is due to VoIP, and how much due to new Skype applications, etc.? I’ll dig around a little and see what I can find, but if someone has a pointer I would appreciate it – Coluccio: Dave O’Leary highlights two interesting points, which I interpret as: The fidelity of Skype’s audio response and the intractability of voice growth. Historically, most voice codecs were designed to fit the bounds of the voice frequency (VF) channel, which have been almost universally limited by telecom industry channel filters, when channel multiplexed line facilities have been used: 2700 HZ of bandwidth on the low side (from 300Hz to 3,000Hz), and 3400 Hz of bandwidth on the high side (300Hz to 3,700Hz). Additional line properties associated with: phase shift, which is the primary factor in analog signal delay distortion, often characterized en masse as envelope delay distortion (as opposed to the latency and jitter we encounter in the IP sense) and uneven frequency-amplitude response (fidelity); AM/PM; and a host of noise contributors further degraded signal quality to the levels that we’ve become accustomed. These line anomalies are non-existent in a world where the encoding and reproduction of auditory signals no longer must negotiate hostile analog line conditions and their attendant constraints, but are enabled, instead, by codecs on either side of the ‘connection,’ by client software. Thus, Skype and others who follow will no longer be bound to the channel’s parameters. Instead, its codecs The COOK Report on Internet May June 2005 can be designed to simply follow the intentions of the designer, following the frequency ranges that they select, and in the process introducing only a marginal penalty when viewing overall packet sizes. of auditing to take place, however, through its logging and IM auditing capabilities, ironically enough. I wonder how those will be viewed by anyone examining them, since no one is actually auditing Skype, itself It was about eight years ago that I engaged a DSP application engineer who had been working on a set of 729/723. a codec designs, in a conversation about stereo and quadraphonic delivery of VoIP for the music and content delivery industries (MoIP?). The TDM equivalents required anywhere from a quarter to three quarters of a T1 system (using multiple, concatenated DS-0s) to produce these results in the past, with the latter typically used for the transport of content from the studios of FM Radio (and TV) Stations to their respective hubs for distribution over the airwaves, which, themselves, were bandwidth constrained, by design. Davis: See below for comparisons with Asterisk and comments of users. Free, open source. Both sipX and Asterix hope to leverage their software to sell add-on products, which has been suggested here as the Skype model. While it was envisaged at the time that such fidelity would someday be available to the content and distribution industries, as replacements for what preceded them in the TDM world, it now appears likely we’ll be seeing such capabilities extended to consumers for ordinary day to day use, as well. Granted, such capabilities already exist in cable TV and other systems that support residential entertainment delivery, but I’m talking here about applications that are under the end user’s direct control and choosing. -- Open-Source PBX Battle Brewing By CAROLYN SCHUK for VOXILLA. COM Tracking voice traffic growth has been tricky business, at best, for a long time now, even before the introduction of VoIP. Enterprises who’ve used tie lines, for example, have tended to obfuscate the matter, as have integrated packet delivery systems supported by Frame Relay and ATM, where the end user organization has dictated what slices of bandwidth would be used for which applications. Since ~1998 or so, the problem has only become much larger for the bean counters, since enterprises, as well as consumers, have been deploying VoIP in their backbones (and from PCs, where the consumer is concerned) in ways that defy auditing from ever taking place. The difference, is sipX is sip only, but easily managed with a web browser. Asterisk has no easy-to-configure tools. Both are non-proprietary, open source. These open source products may be another play on the techie side of the market. http://voxilla.com/voxstory136.html In the open source PBX world, Asterisk is king, but it’s no longer the only game in town. SIPfoundry, a Westborough, MA-based group formed in March 2004 to promote open-source telephony applications based on the Session Initiated Protocol (SIP), has released a complete SIP PBX, called sipX, based largely on code contributed by IP telephony vendor Pingtel. And though Digium, the company responsible for developing the popular Asterisk PBX, has enjoyed a virtual free ride in the world of open source telephony since Asterisk was made public in 1999, CEO Mark Spencer is welcoming the competition. “It’s beneficial to have several companies offering open source systems,” said Spencer. “It validates the concept.” Both Asterisk and sipX are written for the Linux operating system, but there are significant architectural differences between the two. Skype actually ‘does’ permit such forms 115 Asterisk’s architecture is designed around a core PBX engine, with applications layered on top to support a number of different protocols, including SIP, H.323, MGCP, IAX (itself an open-source protocol written by Digium), and potentially other yet-undeveloped protocols. This approach, says Spencer “offers significant benefits‚ for organizations that want to support a heterogeneous network out of the box.” sipX is built from the ground up using 100-percent SIP-compliant blocks. Unlike Asterisk, sipX speaks SIP and only SIP, the protocol which has become the de facto industry standard and is credited for providing the basic foundation upon which much of the VoIP industry now sits. By relying on SIP, says Pingtel CEO William Rich, sipX “provides a very high level of interoperability and scalability.” The architecture of sipX, says Rich, allows systems developers to perform different functions on different servers, replacing the included components with custom-developed ones, or combining them in different ways. [Snip] Stastny: Or hack your Linksys WRT54G Router to be a SER or Asterisk PBX ;-) http://www.toyz.org/mrblog/arsee: chives/00000186.html and http://eurovoip.blogspot.com/2005/02/100-telco. html Coluccio (in answer to Richard Stastny): Thank you Richard. I pulled on the string your (toys.org) link and landed here: h t t p : / / w w w. t o y z . o r g / m r b l o g / a r chives/00000185.html The message on that page reminded me that ordinary end users’ machines can be used as Skype servers. I have, on about three or four occasions, experienced a premature session tear down (on hook from the other end) while talking over Skype. Could those have been caused by end users powering their units down, COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA thus removing themselves from my established paths? While the Skype license states that your computer may be used by others as a server, there is no compelling reason why end users can’t power their units down (or experience crashes, for that matter, or reboots) at any point in time. Any thoughts on this? Another quirk, which I now believe to be a trending pattern, has to do with initial session initiation time. Upon first Skyping someone on any given day, I find that it’s taking longer for the session to initialize now than it did a month ago. A little over a month ago I didn’t notice any appreciable delays in setting up sessions, maybe a few seconds. Two weeks ago it was taking me perhaps five seconds or more. Yesterday I clocked one attempt to reach Gordon at 13 seconds. At first it occurred to me that this might be a scaling problem, due to the enormous numbers of new users coming on line. But further thought on the matter, given the nature of p2p being what it is, I now am tending to think otherwise. Has anyone else noticed these longer session initiation times aside from myself? Curious. And if so, what do you suppose it’s due to? I should add that when this occurs I do NOT have a similar problem re- connecting to the same party if I choose to disconnect and re-connect immediately. nobody said geeks know how to name products). Cisco’s (CSCO:Nasdaq - news - research) router chief has taken a board position at closely held VoIP upstart . Mike Volpi, head of the Cisco technology unit focusing on boxes that direct Internet traffic, has joined the board of the Luxembourg Net calling company, according to a story published in BusinessWeek on Wednesday.” Jennings: I’ve seen several different proxies running on hacked Linksys box. Cool hack, but who cares. We have a name for phones system inside your house that can’t call out -”intercom”. You still need the IP phones. As soon as you want PSTN connectivity, in or out, you start talking about a telephony service provider TSP of some type. Now the TSP is willing to provide the proxy for no more than just the PSTN connectivity so it’s hard to see the incentive for the local box that does call control. I’ll point out people like Vonage charge *more* if you try and connect something like this up to them. Then Toni Li returns to Cisco in December (or was it late November Toni?) to work for mike Volpi. So far so good – now take a look at this. http://www. toyz.org/mrblog/ Feb 22 If it was cheaper to get a ISDN BRI line from the LEC than to get phone service from a TSP, and if the linksys box had an ISDN port, this would be more interesting. “It’s here, the user owned phone system The rebels are at it again. This time they’ve built a free, open-source VoIP platform for embedded devices. It’s a VoIP PBX in box, a cheap affordable box. The box, in fact, is a common Linksys router. It’s called SIPatH (hey In the meantime, keep letting me know about hacked Linksys boxes that do cool VoIP stuff. I love it - very nice platforms to experiment around on. Of course a single intercom system that went to my house, your houses, and everyone else house would be interesting. COOK Report: Check out this September 15 2004 announcement http://www.thestreet.com/tech/scottmoritz/10183131. html 116 Symposium Discussion February 25 - March 10, 2005 VoIP Economic, Quality and Network Traffic Issues Highlights VoIP Adoption Curves Raj Sharma: I am trying to model the growth of VoIP end points and I am assuming that the adoption of VoIP technology for the end points will follow Gompertz or Fisher-Pry models. So, let’s take the Gompertz model – the speed of any technology adoption (the slope of the Gompertz curve) depends on the ‘b’ value. The ‘b’ value for different technology adoptions in the past is given in the following table Analogy Range b Radio 1922-1940 .18 Color TV 1955-1992 .18 Television 1946-1960 .32 CD Player 1986-1994 .17 VCR 1979-1994 .23 Pay Cable 1973-1981 .21 Personal Computers 1980-1998 .10 Cellular 1984-1999 .17 Online/Internet 1992-1998 .20 Average .20 So, I took the average of .20 for adoption of VoIP in end points and came up with an estimate that it will take approximately 6-8 years for majority (80%-85%) of end points to be VoIP. The way Skype is being adopted, this may even seem reasonable. So, my question is, are my assumptions correct? Does the Gompertz model really apply here? Or is this more of a viral phenomena that follows a different adoption curve. Is the implication of this hypothesis that majority of traffic will be VoIP to VoIP in 8 years with very little being originated and terminated on the good old PSTN? Odlyzko: The Gompertz model seems appropriate, and with the parameter you mention, as long as, the choices are made by individuals. What might change things is presence of what I called “forcing agents” in a paper from 1997 that debunked the notion of “Internet time,” http://www.dtc.umn.edu/~odlyzko/doc/ slow.evolution.pdf Forcing agents can push thing along, say when a telco moves its customers, whether they want to or not, to VoIP. Schulzrinne: However, in the absence of forced upgrades, I have some doubts as to whether using the average ‘b’ across the earlier technologies mentioned is reasonable. Unlike almost all of the technologies mentioned in the comparison, VoIP does not currently offer fundamentally new capabilities to consumers. (Presence is nice, but only matters if most of your calling destinations are on IP, too. Better voice quality is nice, but I doubt that many non-techie people would consider this sufficiently interesting to ditch their phone number.) Hughes: When one who lives in Thame, Nepal, a Sherpa village which has NO Nepalese PTT PSTN service at any price, and now gets reasonable quality voice both IP to IP and to Out, that offers new and revolutionary capabilities for ‘would 117 be consumers’ not just incremental or marginally better PSTN service. If you are thinking about only urban US and other dense population cities with excellent telephone services, sure, VOIP isn’t really ‘needed’ except for a handful of people like me who DO make lots of long distance calls and therefore have Vonage at $24 a month (which cut my AT&T bill by $400 a month LD charges) Schulzrinne: I don’t think we’re disagreeing, just answering different questions. I would be the last one to dispute the usefulness of VoIP, but I’m just trying to be realistic when people try to estimate the speed of displacement of existing residential technologies by extrapolating from other (US) technology adoption curves. There is no doubt that VoIP is today very useful as a “niche” residential technology in the US and as a core technology in developing countries. But it’s a big jump from supplying the roughly at most 5% of US households that don’t have traditional landline, bought at whatever inflated or subsidized prices, to 80% market penetration. I think there is a core message that we both agree on: It is absolute foolishness to expend another tax (= USF) dollar on running narrowband phone service to hard-to-serve rural areas. Btw, even the Cisco phone can be configured for G.729 (which has much lower bandwidth needs than G.711), but I think Cisco has been primarily after the enterprise market where this matters less. Hughes: Yes, I even tried the G.729 lower bandwidth over Cisco 7960 SIP phones, but it still failed at 64kbps satellite bandwidth. Only Skype worked to Namche, Nepal. (and with 1,200ms latency!)I hate to kick a gift horse in the mouth on this list, for it was Jim Forster and Cullen Jennings, both from Cisco who got the Sherpa’s set up through me COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA with sets of those Cisco IP phones as a donation from Cisco. So they will keep them around until the price of 128kbps satellite IP comes down from the $1,200 they were paying, even when no climbers or trekkers are around. Cause they really are nice phone instruments with built in speaker, while Skype still requires a PC, and external speaker or microphone. And the Sherpas sure are not very techie. Earlier Schulzrinne: Radio, TV and Internet did. Thus, this is still mostly a price decision for consumers, with some risk (reliability, work during power outages, 911 issues) and inconvenience (wire up ATAs or restricted to PC). Except for the small minority that make tons of domestic long-distance calls, $25-$40/month just for long-distance isn’t a terribly good deal, particularly since many of these long distance calls are now placed over cell phones. Average residential interstate toll minutes per month is 41, i.e., $2 unless you’re too lazy to change your 1975 AT&T calling plan. Total toll minutes is 90 per month, i.e., $4.50 at competitive consumer rates. I don’t have the complete distribution, but since the median is presumably less, getting close to 50% replacement is going to be tough. Data from http://www.fcc.gov/Bureaus/ Common_Carrier/Reports/FCC-State_ Link/IAD/trend504.pdf Vonage and kin obviously make sense if you ditch your regular local phone service, but that’s more of a risk and hassle for the typical household with a bunch of phones spread throughout the home. (I don’t know the details, but my impression is that X.25, frame relay and Telex were displaced by another version of forced upgrades: prices that became ever less competitive or were even raised.) Odlyzko: These are reasonable objections to my comment (and Raj Sharma’s assumptions). We’ll see. But a couple of minor corrections: The long distance usage figures in Henning’s message are way too low. I am swamped, so can’t search for all of them, but just looking at interstate switched access minutes in Table 10.1 of the FCC report that Henning cites, we find (for all of 2003) 444.1 billion, which, coupled with 183 million local lines gives 202 minutes per line per month. Schulzrinne: This is indeed peculiar; it should be noted, however, that access minutes apparently get counted once at each end (see explanation on pg. 10-1) and include 800# calls, so the difference may not be quite as large. It is interesting to observe that long distance volume has gone down significantly, despite falling prices, although I can’t tell how mobile calls count in this calculation. Odlyzko: So the prices are higher than cited, and in practice considerably higher, since most people are lazy, and then there are also all those fixed monthly charges. Schulzrinne: The subscribers to 30cent per minute plans are probably least likely to have broadband Most of them (subscriber line charge and the like) would not be affected if somebody were to replace MCI long distance by Vonage, for example. Odlyzko: Another little comment: I am not sure Frame Relay has been displaced yet. As of a couple of years ago, industry revenues were climbing (and carriers were making money). I imagine by now there has been some cannibalization by the Internet, but don’t have recent statistics. Sharma: A new technology is adopted because it is more economical or technologically superior to the ‘existing way of doing things’. The latter is, as often as not, an identifiable older technology. Thus, the process is referred to as one of ‘technology substitution’ of new for old technology. Regardless of its ultimate superiority, when a new technology is introduced to the market, it is usually expensive, unfamiliar, unproven, and imperfect. The old technology has already achieved economies of scale and is wellknown, proven and mature. In short, the old technology is superior to the new from a ‘business’ perspective. Add to 118 this the fact that most people’s economic interests are tied to the old technology. Thus, at first, the adoption of new technology proceeds slowly relative to the total market, which explains the early flat part of Gompertz’s S-shaped curve. Usually, the new technology finds niche application where its special features outweigh its disadvantages. The first application of electrical lighting was on ships, where fire caused by oil lighting was a constant fear. The first applications of transistors were in bomb fuses and hearing aids, where the package size and low power requirements were crucial. Perhaps ‘presence’ is the first application of VoIP and success in such early applications combined with lower cost is allowing VoIP to pull itself up by its bootstraps. With Regard to Architecture and Economics all VoIP is not Alike Davis: But, Raj, VoIP isn’t so new, and the collapse of PSTN traffic prices, much of which inter-city traffic, along with mobile telephony, is carried in VoIP, doesn’t make it cheaper. For infrastructure not engineered to the psycho-physiological requirements and expectations of voice, VoIP can be, and frequently is a worse product and very much more expensive, given the cost in labor, planning, time, fixing, re-engineering. Legacy PBX replacement will be a move reason when that is necessary. Edge-toEdge is VoIP within same provider, or when encapsulated in Sonet which is trans-provider interoperable out of the gate. I think this one is difficult. I can see/hear the difference with digital cable and satellite on my HD TV and my excellent home theatre system (B&W speakers rendering concert halls and movie theatres rather banal for audio). When Vonage works, which so far has been a hundred percent, I notice no difference. I The COOK Report on Internet May June 2005 know what to expect in terms of quality from the PSTN. I use a different standard with my Verizon CDMA and my Cingular Blackberry. fice aggregation NE? Must the telephone handset itself contain the VoIP client code in order for it to qualify as a VoIP end point? The ILEC’s don’t want their own PSTN. I don’t know what the war cries are about. O’Leary: These are good questions. We are involved with a bunch of VoIP trunking projects with carriers (as described in other messages) where nothing changes at the handset (or with the service model), the carrier is just “converging” their backbone. Sharma: Originally, the question was ‘can you apply Gompertz curve to adoption of VoIP end points?’ And, if so how long will it take to reach 80%+ adoption for VoIP end points. Gompertz curve is an S-curve and the adoption of any ‘new’ technology (I know VoIP is not new technology, but that is relative) takes the longest for the first 10%-15% adoption. After the first 15% adoption there is a rapid adoption to saturation (85%+ adoption being defined as saturation), and the problem at hand is to estimate how long will it take to go from 15% adoption to 85% adoption. Now I know that we may not have reached 15% adoption for VoIP, but the way Skype is going, this may not be too far. So the estimate was that it will take 6-8 years after VoIP reaches 15% adoption for end points to reach 85% adoption. The reasons for adoption are almost irrelevant - in fact, historically, in a lot of cases, the reasons are not even economics or quality - they could be a totally different niche application as in the case of electricity or transistors. So the question is 6-8 years a reasonable estimate given the above? Coluccio: To add some color to the discussion, radio was twenty years or older before its penetration began to make a difference, and TV about fifteen. What would you say VoIP was? If we use the NetSpeak client software launch as a start point, then that would make it what, nine or ten years and counting? Raj, please define what you mean by a VoIP end point, since that seems to be the point of interest here? Does Vonage count, when the telephone set used is an analog device, hence an analog end point? How about a regular POTS phone, where the IP to metallic-copper analog conversion is done in the basement? Or out in the field in a pedestal? or in a central of- On a somewhat related note, and commenting on another parallel thread with the discussion of transition from Frame Relay to IP/MPLS based services, there are service providers that are using the same model as with VoIP trunking. Frame relay continues as the service provided to the customer, and the backbone is transitioned from ATM (or a proprietary technology) to MPLS in the backbone. So measuring bits carried by frame relay vs. IP vs. MPLS rather than revenues derived from the services can generate quite different views of the market. Coluccio: That’s a great analogy, Dave, which captures the essence of what I was saying about VoIP end points. To suggest that Vonage qualifies as a VoIP end point, even though it’s extended to the user over a ten foot analog cord from an ATA, and that VoIP from the basement or field node doesn’t, because the analog cords on those are 500 ft or 1,500 ft, respectvely, doesn’t make any sense to me. Unless a user has direct control of the VoIP client software (application) and can manipulate its varied features in a converged data-voice manner, it’s not a VoIP end point, in my opinion. In other words, the VoIP client code must be embedded in the end user’s device, or as close as possible to it, and must be under the direct control of the user. Other thoughts on this, anyone? Sharma: For call origination, I would define a VoIP end point as any end point from which the originating media gets converted to IP with the use of Codecs before it ever hits a Class 5 switch if at all it does. For call termination I would define a VoIP end point as any end point where IP media terminates without going 119 through a Class 5 switch. I do not count all the Class 4 TDM to VoIP conversion ports as end points. So yes, a subscriber for Vonage, Lingo count as a VoIP end point. Matson: Raj is right as far as history is concerned but in the case of a highly regulated sector like telecoms, “technology substitution” is paced by vested interests and NOT the market of end users. The world is currently experiencing the confusion and turmoil caused by the deployment of advanced digital technologies over the past two decades by the telcos to shore-up their obsolete business models. February 28, Coluccio: Metrics are hard to come by with respect to actual packet counts on FR networks, as by now I’m certain you don’t need to be told. One metric that is both easier to track, and actually makes a far more significant difference to service providers, however, is the amount of revenues that each brings to their top lines. According to a report released in BCR last month, FR revs are still about six times those of comparably provisioned (similar traffic bearing) IP services today, with 2007-8 viewed as the time when they will be more or less the same, with IP taking off beyond that point, leaving FR eventually far behind. Given the radical disparity in per-megabit pricing between FR and IP, however, even knowing the revenue spread between the two is not really a good indicator of how their actual traffic volumes compare. And too, many enterprise intranets use a blend of both FR and IP, often with the integration of the two taking place on the customers’ own premises, in order to circumvent the high costs associated with FR’s predominantly hub and spoke design. More on Muni Networks (Texas and Colorado) Februrary 24 Coluccio: This may appear off topic at this stage of the discussion, or maybe time-shifted by a month or two from when municipal networks were being discussed here. I had to share COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA these stories here, just the same. The urls below will refer you to two municipal fiber optic access networks that are being rolled out abroad: Brazil: http://www.siliconinvestor.com/ readmsg.aspx?msgid=21074902 China: http://www.siliconinvestor.com/ readmsg.aspx?msgid=21074927 see also Lessig essay at From Wired Magazine, available online at: http:// www.wired.com/wired/archive/13.03/ view.html?pg=5 Hughes: I am supposed to be a 45 minute presenter next Friday, March 4th, in Colorado at a “City/County Management Association’s “Emerging Wireless Technologies: Passing Fads or Permanent Solutions Forum” 50 City and County IT types or Administrators will be there. About 4 ‘City’ Case Studies will be presented also, and a panel discussion. Most the cities attending are under 150,000 I am supposed to talk about - ‘future wireless technologies’ (which, of course, might be applicable to City or County Governments What wireless technologies on the horizon do YOU think would be of real value to the operations of local governments like these? Cheponis: Two main technologies need exploring: meshed 802.11n (or g or b) and Wi-Max. I don’t believe the Wi-Max hype, but it’ll no doubt be useful in some situations. Wi-Fi is here to stay, it’s the Ethernet of the Airwaves. Coluccio: The merits of WiFi, and the questions surrounding WiMAX, notwithstanding, Dave, you may want to think about some of the enabling and ancillary wireless technologies that will foster backhaul in some cases, and direct high speed access to schools, municipal buildings, libraries and hospitals, etc., in other cases. Such as some of the millimeter radios like those being supplied by Terabeam and Gigabeam at this time that will support gigabit speeds (slated for 10Gb/s soon). When used in concert with fiber in a hybrid fiber/wireless configuration, these millimeter radios will also support the kind of wireless mesh topologies in some rural areas that would otherwise be impractical or too expensive, even where fiber from incumbent providers already exists. Yokubaitis: Lucky Colorado Municipal entities that can still build such projects. I just got back after lunch from the Texas Legislature where there is filed a new Telecom Bill (HB 789) ( deregulating Incumbent Tel cos). HB 789 includes language that prohibits a municipality from building or selling telecommunications services OR partnering with any organization ( wireless service provider) that does. The “partnering” part may go away before a final version surfaces for a House vote. But the part about the city owning, operating or selling telecommunications (wireless) services other than for internal municipal services will probably stay. “Vamos ver.” Cities are up in arms. Texas Municipal League of Cities is roused. Private Community Wireless initiatives are up in arms. The Texas Legislature, or at least members on the Regulatory Committee holding hearings on the House Bill 789, does not want “tax money and municipalities competing with private money and industry.” http://www.house.state.tx.us/fx/av/ committee79/50223p35.ram Hertz: I know you asked about technology issues, but having dealt with, face-toface, dozens and dozens of communities (POP 2k to 90k) and their so-called IT departments over the last 8 months, technology really isn’t the issue. The people that make the decisions have the 50,000foot view and don’t want to get any closer. The IT people, while not clueless, tend to be very narrowly focused; they take the 1-foot view, and don’t want to look up. As a result there isn’t much shared understanding between the two groups. If you are speaking to IT people, then maybe the best bet is demystification. A history lesson may be appropriate, with history being more than the last 10 years. After all, the common use of the term “wireless” pre-dates “radio.” People have been using wireless for more than 100 years, so we certainly aren’t talking about fads or new discoveries. There is nothing new about the physics (disregarding nano-small and cosmo-large), and there are certain physical law limits (same caveat) that apply regardless of whose technology implementation you choose. By grounding the IT people, they can better connect with the people that will make the decisions to invest (or not). And like most management decisions, it isn’t the best technology that wins. Who does? However, small towns across this vast homeland of ours (Texas) are desperate waiting for the promise of broadband to finally arrive via entrenched Incumbent telcos. Unfortunately, many areas of Texas are closer to remote parts of the Third World. There is no economic counterpoise to the juggernaut Incumbent except the municipal governments. They want access to something other than high cost dial up i.e. Broadband, that they have been hearing about for years. Forster: I think at this stage there’s no need talk about the technology at a City/ County Management Association, or at least no need to bring up an alphabet soup of protocols. You can leave that for a separate discussion with the technical people. I think you ought to talk about the role of government or other quasipublic bodies — merchants associations, homeowner associations, etc. , and how they don’t need to become ISP’s per se’, but can still play a major role in physical layer through siting, permits, etc., and potentially in the financing. Coluccio: Here’s the full-length video on Wednesday’s hearings in Texas: Editor’s Note: at http://wifinetnews. com/archives/004870.html on Feb 21 120 The COOK Report on Internet May June 2005 Chuck Larrieu asked: Tell us why is muni wireless a good idea? Good answers do not include “because it’s kewl”, “because I think its a good idea”, “because the poor need it” and other selfish or pretentiously altruistic responses. The utility argument is not a good answer either, because the response is “why not privatize utilities?” Another response is “why not socialize other private businesses that government deems important to the economy or beneficial to the poor?” Hughes: Here is my answer to the perfectly reasonable question Larrieu posed on Glenn Fleischman’s forum. I have found that a woman Colorado Senate legislator, Jennifer Viega, Democrat no less, has sponsored SB-05-152 which would all but block any Colorado town from setting up their own municipal wireless (or any other type) public network. The Colorado Municipal League is against that Bill, for good reason. Consequently, I am not only going to educate the 50 Colorado Town/County Government reps at the Wireless Conference next Friday, I am going to yank the bell of every Colorado Springs Senator and House member, whether Republican or Democrat and get them to kill that stupid bill. “There’s one damned good reason, Chuck, why NO State Legislature should be allowed to block Municipal Wireless networks. And that is because the Telephone Companies have TOTALLY FAILED to bring Broadband services to every town in the US, even though they are guaranteed rate-of-return monopolies at the local loop level! And even though ‘Universal IP Access’ is the Policy of the United States Government! AND they charge Universal Service Fund Access fees out of YOUR pocketbook, AND if they had any brains they could extend IP wireless to EVERY farmhouse in spread out Colorado for PEANUTS, and charge $50 a month for every remote customer who would be glad to pay it. I have been actively engaged in bringing broadband to the most Rural areas of America, including the most difficult and poorest areas of huge Colorado from the day unlicensed wireless was FCC authorized and the first radios came out - about 1990. Everywhere the story has been the same - the small and rural towns beg the Telcos to bring broadband to them for their economic and educational future. The Telcos have dragged their feet, tried to obstruct any one ELSE bringing broadband, and then when they have been forced to have charged obscene rates ($2,000 a month for a T-1 forty road miles from Alamosa to San Luis School. While *I* put up a pair of Unlicensed radios that cost less than $2000 each and got the same thing 30 miles FREE between the radios. In 1995! Where wireless is concerned, the RBOCs are not only greedy, and too stupid to see the coming market, they are also more often than not incompetent. I hope the Muni-Wireless bandwagon really gets rolling, and for the first time REALLY give the State-Anointed Telcos competition. That will get them off their ass faster than ANY anti-muni legislation. And I frankly don’t give a damn if municipalities are in competition with private Wireless ISPs, because very very few Wireless ISPs can on their own connect up complete towns and cities. They don’t have the capital, the scale, or the highly profitable marketplace that is needed before any Venture Capitalist will part with HIS money. And I AM a wireless ISP in Colorado Springs. I’d LOVE for the City of Colorado Springs to do Municipal Wireless; I’ll have more business extending wireless to places no city wireless network is going to go and make more money if they DO than if they DON’T. Make the Internet Pie Bigger, don’t fight over the Wireless pieces. If ANY of the large telephone companies were real risk-taking profit or loss private businesses, and did NOT have government filling their pockets by regulation, there might be a case for holding back muni’s. But they are NOT real businesses. They are fat slothful Dinosaurs who require governments to stay in business. Make the damned Telcos compete for 121 a change. Or run their ass out of town when they go bankrupt. I have my Voice over IP, both with SIP, , and can connect up to a SATELLITE IP feed any time I need it. I don’t need Qwest for anything anymore. They had their chance when the Internet first came in. They failed then. They had their chance to USE cost effective wireless for the last 12 years. They failed at that too. They had their chance with Voice over IP. They still are failing compared with ANY private SIP VOIP company. Now they deserve to die. That’s called survival of the fittest where I come from, even if a hell of a lot of Republicans say they believe it, but then feed at the government trough themselves. And if Municipalities deploying Wireless can be their Pall Bearers, so much the better. The Muni’s will learn soon enough its not as easy as it seems, and be happy, in many cases to invite in private Wireless companies, subsidize them with tax breaks, red-tape cutting, and support just like they do all OTHER job producing businesses. Coluccio: Dave, Just in case you’re not angry enough, read this one, where Bell South blackmails Louisiana to kill muni fiber project: h t t p : / / w w w. t e c h d i r t . c o m / a r t i cles/20050225/1232223_F.shtml On Wireless for Voice and Interactive Applications – QoS OpEx and CapEx Implications On February 25 Roberts: We have tested WiFi (802.11b and g) for many months to so if we could use it for Skype or other VoIP to PDA’s. The problems with voice and 802.11 are severe in a loaded environment. 802.11 is based on slotted Aloha (which I invented in 1971) to request a transmission slot. If there is other competing traffic, the transmitter waits and tries again. There is an exponential backoff technique so that one may be delayed for more than most voice systems (including Skype) can tol- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA erate without blackout periods (drops). This makes 802.11 almost useless in a heavy use area. I would hope that load control could fix this but I have no proof yet. I do know that at home and at work, 802.11 (either b or g) is too bad for me to use wireless voice over it. If instead, I use a wired computer, and Skype works perfectly. I expect the same problem was the problem some of you were having on cable where congestion grows each month. Note that at first, at low congestion, 802.11 for voice works great. So many people will then install it, later finding out that data works well, but voice is hopeless. On the other hand, no matter how much I dislike the central design of Wi-Max or cellular, both have been designed to support highly interactive data (voice, etc) with low delay variance once the call is established. Thus WiMax, used in open spectrum, may in fact have some future value. WiFi (802.11) might be fixed with some protocol changes. It will not happen with the currently deployed cheep boxes but smart vendors could perhaps fix the delay variance problem. These points are not well understood yet, but it is worth warning peopleabout them. Coluccio: I want to call attention to article in the January 10 issue of Network World “Voice over Wireless LAN” by David Newman http://www.nwfusion. com/reviews/2005/011005rev.html. The author took part in extensive tests using various vendor’s access points, with and without QoS. Without QoS being invoked, and traffic levels being constant for all tests performed, only one or two talkers could coexist in the same zone at quality levels that were livable. Beyond that number quality deteriorated rapidly. With QoS invoked, the number increased to, I believe, 5 or 6 users. Unfortunately this article is not posted to the BCR site yet, but if you can get your hands on the hard copy it’s worth the read, in my opinion. Southworth: This is an article quoting my business partner that I think is appropriate to our discussions http://www.vonmag.com/issue/2005/feb/ VoIP/Video Quality of Service, by Richard Grigonis Size doesn‚t matter– at least for service providers carrying multiplayer online games. OK, it matters a little, but in the scheme of the “my pipe’s bigger than your pipe” trash talk that‚s transpired between cable and telcos for years, size is practically irrelevant. All you need is broadband. Quality of service, on the other hand, does matter. That‚s why the cable and telco industries are racing to see who can deliver data with low latency, no jitter and no delay. The one with the most reliable connection will likely be the de facto carrier of choice for what’s now a niche market of gamers that‚s expected to grow exponentially in size and value this year. To be clear, today’s broadband pipe is adequate for interactive gaming and it‚s being used today without much worry. Big-name gaming companies like Sony and Microsoft already „have our own infrastructure; it’s open to all broadband companies” according to Sony Computer Entertainment spokesman Ryan Bowling. And that’s enough for today’s gamers. But best-effort broadband connections are less than optimal; real-time interactive gaming demands a level of control that exceeds even the most stringent VoIP requirements. “For voice-over-IP you have to have the round trip time be 300 milliseconds or less. Gamers are looking for about a 30-millisecond latency; something 10 times better than what we’re doing for voice-over-IP,” says Peter Macaualey, an independent consultant from Reston, Virginia, who serves on the DSL Forum’s technical committee. The DSL Forum is upgrading its VoIP QoS specifications to meet what gamers 122 need. While DSL – sans fiber-to-thepremise – can’t match the size of cable‚s broadband pipe, it can level the playing field by delivering low latency QoS with benefits that go well beyond just letting two gamers have at it. “This is something for the business environment that would make the voiceover-IP experience better” Macauley says. “We need to make a business case for quality of service and we can do that in a business environment pretty easily.” Cable maintains a nice lead in high-speed data networking but can‚t rest on its laurels because the telcos, with a platform more widely viewed as data-friendly, are creeping up. So cable is steadily completing details on a next-level DOCSIS 1.1 specification– PacketCable Multimedia (PCMM) – that covers a variety of needs, including setting rules for how operators can segment their broadband networks to make room for dedicated gamers. [SNIP] . . . . QoS OpEx Economics - Flows or UCLP? Bill St. Arnaud: Technically this [what the BCR piece above says about QoS and OpEx] may be true, but rarely do you see a systemic comparison of QoS and OPEX versus an alternative of increased CAPEX to significantly increase the bandwidth. The OPEX costs of managing QoS are horrendous and, because they are so labor intensive are only going to increase. CAPEX costs per bit of bandwidth continue to drop dramatically. Many carriers feel they will be a charge a premium for QoS that will more than offset the OPEX costs. But, it is hard to imagine how carriers will recover a premium for a service like voice that has never seen as a premium service by the public. Can you imagine carriers trying to sell a service that ensures greater probability of having better sound quality or a higher likelihood of not being dropped? Odlyzko: Very well put. QoS introduces new levels of complexity into an already complex system. The COOK Report on Internet May June 2005 Roberts (to Bill St. Arnaud): I’m not sure what I said that triggered your comments about QoS Capex and Opex. However, I am happy to comment on the issue. However, to clarify what my comment was about, it was that Wi-Fi or 802.11 has serious problems with delay variation in practical use. I only had same small private experiments but the paper referenced by Henning Schulzrinne confirms my concern, only a few (10-30) VoIP calls are possible over 802.11b and with normal computer FTP, very few will work acceptably. 802.11g would reduce the delay, but has a much shorter range and thus coverage. This should be of great concern for cities that are installing 802.11 across the city if they think it will support WiFi phones. The same holds true for Motorola and Skype if they hope to use cheap phones using 802.11. Now, back to your economic point, Bill. I now believe that routers built to route flows rather than routing every packet, eliminate a vast amount of complexity, reflected both in CAPEX and OPEX. When flows are routed, it is easy to manage the QoS of the flow. The more the flow is like a fixed rate TDM flow the easier is it is to manage and support low loss and low delay variation. The more a flow is like TCP with big bursts and uncontrolled growth, the harder it is to support and manage. Perhaps the higher cost of managing TCP over VoIP can be offset by the use of cheaper excess bandwidth, but overall, as you know, fiber bandwidth is the lesser of the costs today. So I would contend, that bit for bit, a constant TDM stream like VoIP should be cheaper than WWW or FTP bits. I cannot predict the charging model that will be used, I am only looking at the CAPEX and OPEX cost. Carriers will charge what the market will bear, and if voice sells at a premium, they may charge a premium. But as flow routers become more widely used and the competitive market settles down, the price per bit may be the same for any QoS. That is because, managing QoS only requires discarding and scheduling correctly, two much easier things than routing and DOS on every packet. Done correctly, there is no increased OPEX to set up or manage QoS. It is a specified by the sender, either in the DiffServ mark or in other packet information. I realize it will take time for me to prove these points, but do not assume that the structure of routers has to remain as it has for 30 years and that with a new design, that the cost and complexity does not need to appear in the same place as before. I know full well you understand this point. So please accept the possibility that QoS can reduce cost, not increase it. I will prove it to you very soon. (Andrew, the same for your comments). I really do worry however, about QoS at the wireless edge, since the 802.11 protocol is imbedded in the wireless devices and cannot be improved easily. People may assume the same performance as on the wired network, and this is not yet the case. St. Arnaud (to Roberts): It is not the technical costs of OPEX that are my concern - it is the human costs of managing SLAs and billing systems that QoS entails. In the wireless example, there is no question that today’s WiFi has limited capacity to support QoS. One solution is to implement QoS so that those customers who have contracted with you to provide a service are guaranteed priority when they make voice calls over your WiFi system. The other approach is to install a lot more WiFi nodes. Assuming everything else being equal, for a given number of Wifi VoIP customers, which would be the cheaper solution? This analogy can be extended to so many other network examples where it is perceived that QoS is the only answer. There are basically two types of QoS systems: internal and external An internal system is one that may be used by a service provider at the edge of a network so that can prioritize traffic within their network - but not offer SLAs to their customers. This is something MSOs want to deploy in order to support triple play on the cable system between the head end and the customer. This makes a lot of sense and is relatively inexpensive An external system is where a customer needs end-to-end QoS and wants a SLA. To my mind this is very costly and expensive because of the SLA management issues regardless of whether you use flows, MPLS with premium service, WRED, TDM SONET or other techniques. However, an alternate approach is to give the OPEX knobs to the customer so that can deploy and manage their own QoS. This gives the customer what they want without incurring a huge OPEX overhead for the carrier. This is what UCLP provides - by using web services we can let the customer control their specific knobs for managing and controlling end-to-end QoS. This will work with flows as well. Roberts: Bill, I agree with all you say here. There are two ways for the customer to set his own QoS, one which you mention is a Web service, and a second is to allow some in-flow QoS signaling, either Diffserv or more generally as I have done in the TIA and ITU QoS signaling specifications. But I agree that billing should be simple flat rate based on the web setup of a profile. PSTN history forced me to do call by call billing when I started Telenet back in 1975 (Sprint today) and the billing cost as much as the service. Thanks for clarifying your point. Coluccio: to St. Arnaud - I agree with your explanation for applications that leverage abundance, as in those cases you’ve described in the past where universities and large enterprises were provided their own lambdas, which I’m sure could commute to other situations, as well, where very high capacity L1/2 links are managed by end users who have access to the web services you’ve mentioned, which, in the case you cite is namely UCLP or equivalent. Does this scale to the average end user sitting in their living room, study or office at home? What I’m about to state holds true for the end sitting behind an enterprise desktop computer as well as the residential user sitting at home. Consider the total personnel-hours/days/ 123 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA years that would be involved, if you’re concerned about human costs, if every user in a large corporation or residential serving area had to adjust for their own QoS by manually setting up macro parameters of their own, and adjusting periodically for trends and shifts. The sum total of all users’ time and labors would be such that outsourcing those burdens to the service provider would make eminent sense. A kind of equilibrium would be reached, similar to what we have today, where the carrier is responsible for not only the channel’s raw capacity, but for governing flow prioritization and packet loss for all services that share the common channel. I don’t want to appear to be nit pickin’ the points you made, for they are valid in the network settings that you’ve described. I am stating, however, that different networking domains demand different forms of attention and treatment, and where you have a residential serving area at the edge (or office complex sitting on the edge) consisting of tens of thousands or millions of subscribers all being fed from the same cookie-cut reference design pipes at the access layer - all of them being too narrow in bandwidth, by nature, to qualify for the benefits afforded by abundance - economies are best achieved by the service provider, and not the end user, performing policy enforcement and QoS. Since I mentioned abundance, I should remind myself, and perhaps others here as well, that abundance is merely a relative term, and what might be designed into a network today as an abundant resource, such as bandwidth, quickly becomes the gating factor in the future as application-creep continues to claim more cross-sectional real estate in the pipeline. Editor’s Note: Meanwhile the optical network progress we wrote about at the beginning of 2005 shows no sign of slowing down. On Tuesday March 15, Bill St. Arnaud posted the following to his Canarie mail list. FYI, 1,000 channel WDM transmission is successfully demonstrated in an in- stalled optical fiber for the first time in the world - 126 km-span transmission trial on the JGN II testbed - is not sufficient to make claims like the preceding one without stating where they apply. h t t p : / / w w w. n t t . c o . j p / n e w s / news05e/0503/050308.html Bill St. A’s assertions in the above give and take, where he stated that the complexities and costs of QoS were best left to the end user while offering the UCLP approach as an example, were correct, with the following caveat. They were only correct if one chose to view large enterprise on their own fiber nets or RENs using lambda based networks, and the “users” in those instances were also the administrators of those larger organizations who had high capacity bandwidth swaths at their disposal. [With demonstrations like this it is not inconceivable to imagine a world where every research institution, if not researcher, would have its own long haul 10 Gbps wavelength. ROADM and ULH techniques may also allow all the active optical elements for wavelength insertion and removal be located at customer premises, while the core of the network consists of solely broadband optical amplifiers and ROADMs. This will allow systems integrators and other organizations to deploy condominium wavelength networks where very participating institution or enterprise can manage its own set of long haul wavelengths. Thanks to Olivier Martin and Harvey Newman for this pointer -- BSA] Hughes: There’s no free technological lunch. And this is true even with free, open source software OR with ‘unlicensed’ wireless, or virtual ‘giveaway’ (extremely low cost) chip sets. Not only do you have increasing complexity, BUT you also have a real, and increasing, TIME cost (labor, which in some quarters = $$$) to master this stuff, make it work, make it interoperate, make it standard, make it compatible. By individuals, by organizations, by schools and universities, by small companies, by big companies, by governments. But Bill’s assertions in this regard tend to lean toward being incorrect, in my opinion when we focus on individual, highly-regimented end users who are only one among millions either at home or at work, as though they were stamped out recipients in the reference design of a larger provider’s (enterprise or carrier) sphere of influence. In the absence of a larger provider’s design, however, I believe Dave’s characterization of chaos ensuing to be correct, especially in triple play circles and where organizations’ VPNs and other enterprise-specific applications come into play, but not necessarily when e-to-e and p-to-p applications are considered, although the more isoochronous-like those tend to be, the more they fit into the former category, as well. At the expense of appearing redundant over this point, for any discussion of this nature to be meaningful, one has to be specific about the network venue, who its tenants are, and which dimension of networking is being discussed. [For those contemplating municiple wireless systems], it will be like a Big Wireless Candy Store in the Sky. Until they try to implement them all in their town, thinking they are free or cheap costing them little in their city budgets. Except time. And THEN begin to see why technology and communications companies exist, as well as licensed radios. Technical Aspects of VoIP Traffic Shaping on Wi-fi Network Coluccio: Dave’s comments while correct in their own right, have caused me to recall the discussion on the cost of QoS being one of network opex and capex costs and, as I asserted, time costs on the part of end users, as well. But it On February 26, David Reed replied to Larry Roberts technical 802.11 question: Larry - just to clarify something you said. The 802.11 standard supports slotted Aloha operation, but it is wrong to say that it *is* slotted Aloha. 802.11 124 The COOK Report on Internet May June 2005 supports both DCF and PCF operation. DCF (distributed control function) is slotted aloha, and suffers the problems you refer to. PCF (point control function) is a centrally controlled polling protocol, with every transmitter getting turns allocated by the so called “access point”. Actual 802.11 networks - the ones you call “cheap”, even the “cheapest” ones - actually use PCF whenever an access point is present. Even the $100 access points use PCF, and tell their clients to use PCF. Schulzrinne: I’m sorry, but this is wrong. All access points today use DCF, with PCF rare to non-existent. As far as we’ve been able to determine, Cisco access points are capable of doing PCF, but we’ve never seen it being used. (Also, DCF uses CSMA/CA, which is related to Aloha, but not the same thing. In Aloha, the sender can’t listen to the channel; in CSMA/CA, it does.) This makes sense since data transfer performance for PCF is dismal since a large fraction of the bandwidth is wasted on polling. Reed: Quoting from the 802.11 Handbook (IEEE), “with proper planning, the PCF is able to deliver a near-isochronous service to the stations on the polling list. ... While the PCF is an optional part of the IEEE 802.11 standard, *every station* is required to be able to respond to the operation of the PCF”. That means, access points are not required, but if an access point is present, every station is required to operate in a polled mode. Schulzrinne: Except that APs don’t actually use it. We have worked extensively in this area recently; see our WCNC paper, for example, for PCF extensions. (http://www.cs.columbia.edu/~hgs/papers/Kawa0503_Improving.pdf) This is just an example, and a rather specialized one; there are many more papers, some cited in that paper, on VoIP capacity for DCF (and a few on PCF). The problem with some PCF-is-better papers is that they assume that the PCF interval and the voice interval are perfectly lined up. Nothing ever is, so you get into the situation where the station misses the poll by just a tad, then sends during the contention period (as called for) and then has nothing to send during the next content-free (polling) period, wasting the poll. Also, with silence suppression, about half of the PCF polls are wasted on temporarily silent stations. Reed: As you certainly know, polled networks are NOT slotted Aloha, and are appropriate for voice. In fact, that is exactly what WiMAX is - a polled network. Schulzrinne: Unfortunately, in practice, the voice capacity of PCF is actually lower than for DCF, at least under the standard voice models we’ve used. Reed: So there is no difference resulting from the Medium Access Protocol between WiFi and WiMax regarding its appropriateness for VoIP (SIP, , ...). Where there are differences, it results from overloading the local access capacity. It is trivially easy to overload a 10 Mb/sec network with a single file transfer. And the situation is worse when your “uplink” (as in a home network) is only a few hundred kb/sec, matched to a 10 Mb/s local access network. Schulzrinne: And SIP works fine in networks where the average load on the bottleneck link is 50% or less, and the biggest packets are a small number of milliseconds long on the bottleneck link. The bottleneck link is rarely the wireless access link, but can be the next link upstream. This isn’t the result of slotted aloha at all, but is instead the result of overloading the upstream link input queues, creating a transient, but slowresolving traffic jam in the buffers on the source side of the upstream link. Nor is it particularly due to wireless vs. wired link reliability. The same problem with voice happens in wired LANs with under provisioned upstream paths, and the same solutions work, but they are not implemented in the LAN equipment in either case. They must be implemented in the endpoints or the router queues, or both. Reed (in response to Schulzrinne above): I’m ok to discover I’m wrong, but I 125 didn’t base my statements only on that book, which was written by long-time chairs of the 802.11 committee, and vetted by implementers at Intersil and other primary implementors. I based them on actual empirical observations carried out by me and a student in exploring the actual MAC layer of installed APs at MIT. I may have even looked at Netgear and Linksys APs at my home in the process. PCF was present in all cases. This was not about VoIP, so I bow to Henning’s experience with VoIP problems on WiFi. I didn’t find a reference to PCF not being implemented in that paper, Henning. I would like to see evidence that access points on the market do not implement PCF at all. I have had graduate students studying the actual frames sent in our access points, and that work (last year) showed PCF present in the access points. So I would be interested in makes, models, and actual frame traces that show pure DCF operation of a commercial access point (say a Linksys). It is a different question whether the particular PCF implementation in an access point supports isochronous traffic well. The phenomenon you describe (of units sending frames into the DCF period of the MAC protocol, instead of waiting until the next PCF) would indeed cause problems due to CSMA/CA issues. Roberts: I want to thank both Henning Schulzrinne and David Reed for helping me understand what the problems with 802.11 for VoIP are and why I have not been able to make it work. Clearly there are settings to optimize but even then FTP in the same zone (which we have lots of) makes delay grow quickly. Also the number of phones is limited far more than the “10 Mbps” would suggest. But from this information, perhaps we can control the traffic to make it work, up to the natural limit. Thanks! And on March 1 David Reed: Henning, et al. - I have spent a few days in my spare time doing exactly what I often suggest others do, checking with my own hands and eyes. Executive summary - you are right, and I was wrong. The vast majority of access points I scanned claim that they don’t implement PCF, so they don’t COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA support a polling mode, and instead are CSMA/CA. [the work last year had me looking at 802.11 beacon frames, but whether PCF was enabled didn’t matter, so I guess I misremembered :-( ] Kismet is a quite reasonable tool for examining the protocols implemented by access points, to see who claims to implement the PCF function on cheap access points, as opposed to DCF. When one wants to pursue the engineering truth (rather than the marketing FUD) about networking technologies, it is very useful to do hands-on exploration. (I remember the many demos of “bad quality over VoIP” by marketroids that came down to the fact that most PCs had problematic sound cards and drivers and bad VoIP software that didn’t understand how to use those devices in low-latency mode). the past couple of days, for which I am grateful to everyone here. Larry, with the exception of the WiFi part of this discussion, when you refer to flows, I take it you mean MPLS flows between the edges of the ‘Net, as defined by the public facing edges of ISPs. Edgeto-edge, in other words, and through the core, so as not to confuse the term ‘edge’ as meaning the customer end node. I follow your logic and agree with it, but at the same time feel compelled to caution that edge to edge activity that is internal to the cloud is only one, albeit a major, part of the end-to-end connection. I have to go back to Bill’s earlier statement to make a point that I feel applies equally to some of your last post as it does to his. Bill, you stated: The mystery of networks to users, and sometimes even “experts”, often leads to treating the somewhat inaccessible phenomena on networks as “magic” confusing correlation with causation, especially if one has a prejudice (wireless is unreliable) that makes the explanation plausible and confirms the prejudice. However, a few minutes (or hours) with Ethereal or Kismet often leads to a different conclusion, if you have the eye for it. I strongly suggest being skeptical of all “experts” (even/especially me! :-) - I suspect that no one is more skeptical of what I think to be true than I am, which is why I went back to check myself). I still believe that the bulk of problems with Skype and VoIP over WiFi arise from overloads at the upstream access link, not with WiFi per se. In any case, sorry for misleading you all. Schulzrinne: Thanks for checking - I wish more people would take the time as you did... Successful Video Accommodation by Microsoftʼs IPT Platform Coluccio on Feb 28: The interchange on this list has been very enjoyable for “... but rarely do you see a systemic comparison of QoS and OPEX versus an alternative of increased CAPEX to significantly increase the bandwidth.” You’re apparently basing your assumptions on the existence of statisticallyfriendly large flows within the Internet proper that support traditional IP service mixes, where the cohabitation of best effort and time-sensitive traffic can arguably be accommodated simply by some degree of inexpensive bandwidth over provisioning. Agreed. Where a large backbone is concerned, and where, as I stated a few messages upstream, where properly designed infrastructure (LANs and intranets, in my earlier examples, but public Internet domains, as well) are used for transport. But it is far easier to generalize about the domains of transport and be correct about over provisioning than to make the same assumptions where local or metro forms of transport and their tributaries, and, especially, where one is focusing on access level connections in the deep outside plant where they meet the customer location. It’s with some hesitation that I go on here, due to the likelihood that I too will wind up generalizing, but just to make sure that I don’t, I’ll state that in the following I’m going to limit my 126 focus to the conditions that exist on last mile residential services today, where socalled broadband services are supported by DSL, Cable Modem and even some forms of FTTP (the bandwidth scarcity perpetuating variety), are used. These exceptions to the qualities enjoyed over backbone flows hold doubly true for the growing number of end users who are now opting for bundled (3P) services and receiving squeezed-down droppings of bandwidth on their last mile connections. Clearly two-thirds of the RBOC’s promised IPTV subscriber connections are going to depend heavily on Microsoft’s coming across with their compression algorithms in a timely manner, in order that those subscribers’ DSL connections are able to actually support all that the carriers are saying they will. All of which means, that all of the i’s and t’ have yet to be dotted and crossed on just how Microsoft’s IPTV platform will accommodate video, yet. And in recognition of this, Verizon has actually stated that their initial roll outs will support analog video over fiber until the IPTV capabilities are fully developed and delivered. The problem isn’t really that onerous for Verizon, yet, because most of their published roll-outs are all-fiber based, to date. Bell South and SBC, on the other hand, have eschewed all-fiber build outs, and are deploying copper end sections using DSL in their FTTP deployments, instead. Those are the ILECs that have something to think about, as Microsoft works to stabilize their compression applications in time for service delivery. And these are the same ILECs, along with many other smaller independent telcos who are delivering 3P over DSL, who will be using QoS mechanisms to the hilt in order to make good on their voice over IP quality assertions, and at the same time support up to three or four simultaneous TV sets in the home, with one of those being HDTV. All this will take place over <20 Mb/s, and sometimes <8 Mb/s, offering fewer simultaneous viewing capabilities, along with multiple voice “connections” and “broadband data” support. Good Luck! The COOK Report on Internet May June 2005 And this is quadruply true where suboptimal forms of DSL are used, and this again assumes that users in this case are receiving combined (3P) services that are all contending for the same meager number of bits. To be specific, carriers are offering triple play services over lines that support anywhere from 1.5M/384k to 24M/1.0M over last- section copper xDSL lines. So, there is not the same friendly mix of conditions on the last mile stretch as exists in the backbone or the ‘Net’s edge-to-edge, which explains why there is an entire cottage industry developing virtually overnight to handle triple play traffic intermediation, on-the-fly engineering, policy enforcement and QoS management at service providers’ hubs and head ends. When you (Bill St. Arnaud) state ... “The OPEX costs of managing QoS are horrendous and, because they are so labour intensive are only going to increase. CAPEX costs per bit of bandwidth continue to drop dramatically. Many carriers feel they will be a charge a premium for QoS that will more than offset the OPEX costs... “ Coluccio: I think prudence would support keeping in mind that, while what you’ve stated is true for the most part, it is predominantly true for backbone economics as time progresses, as is being evidenced by the take up rates of last mile triple play services, and the types of facilities and architectures over which those services are being offered. St. Arnaud: “But, it is hard to imagine how carriers will recover a premium for a service like voice that has never seen as a premium service by the public. Can you imagine carriers trying to sell a service that ensures greater probability of having better sound quality or a higher likelihood of not being dropped?” Voice, Caspian, Broadband and Korea Coluccio: Voice is a funny animal. When offered alone, it’s hard to make a buck on it. Today, if it is offered as an add-on to video services, whatever rates are used go almost entirely to the bottom line of the provider, if the service is being priced competitively with the ILEC’s or another contending triple play provider’s. The same holds true for Internet access services that they, and this is true because of the huge economies that exist when the all three services are offered over a common infrastructure. If the video component of the bundle can carry the weight of the build out (capex, and ongoing opex for the majority of last mile provisions) and bring in a few dollars per sub, then the other two services of voice and Internet access are very often pure gravy to the provider. Larry, I guess I can now retract my question regarding flows, since I was late to recall your earlier explanations of Caspian’s approach, which was refreshed in my mind by the release I am posting below. Congratulations! From ConvergeDigest.com: Caspian and ETRI to Develop Flow State Routing for Korea’s BcN 28-Feb-05 Caspian Networks has been selected by South Korea’s Electronics and Telecommunications Research Institute (ETRI) to co-develop advanced IP flow state solutions for the nation’s Broadband convergence Network (BcN). The project will combine Caspian‚s flow state technology with ETRI‚s technical expertise in network control and simplified multi-protocol labeling switching (sMPLS). The BcN project is targeting a goal to build an integrated network in South Korea with bandwidth of 50 ~ 100 Mbps per end user that can offer seamless multimedia services to 20 million Korean subscribers using wired and wireless communications among heterogeneous networks. South Korea already has the highest broadband penetration rate in the world and high-speed residential connections offering tens of megabits of performance have been on the market for some time. The BcN initiative aims to drive connectivity to the next level, offering seamlessly connectivity between services. Network providers benefit with 127 deterministic, guaranteed IP Quality of Service (QoS) offerings, mobility, and security for new and demanding multimedia applications as set forth by the BcN project. Users benefit with cutting edge integrated services, like IP video and music, online entertainment, realtime health and welfare response, Voice over IP, and more, regardless of device or location. Caspian Networks, which is based in San Jose, California, has been developing flow-based routing since its founding in 2002 by Dr. Larry Roberts. In flowbased networking, packets are routed as whole flows, i.e. streams of related packets, rather than as individual packets as in current IP/MPLS networks. Caspian’s existing “Apeiro” platform examines each packet entering the router, identifies flows, and then stores to memory the flow‚s relevant routing information as well as its QoS, loss, delay and jitter characteristics. Flows are identified by the combination of source address, destination address, source port, destination port, and protocol. Subsequent packets in the flow are switched based on the “flow state” data already in memory. By tracking potentially tens of millions of microflows per 10 Gbps interface per second in hardware, Caspian said its Apeiro platform provides deterministic QoS for premium IP traffic that is equivalent to ATM. The ASIC-driven platform is capable of handling flow set-ups significantly faster than the circuit set-up rates typical of ATM and MPLS. Caspian will establish an R&D center in Korea to support the co- development efforts with ETRI and to serve as a technical support hub for the Asia Pacific region. Initially, Caspian expects to hire about 15 engineers. http://www.caspian. com and http://www.etri.re.kr 28-Feb-05 * In Sept. 2004, the government of South Korea picked 3 consortia to conduct Broadband Convergence Network (BCN) trial operations scheduled for this year. The consortia are led by led by Korea Telecom, SK Telecom and Dacom. The goal of the BcN is to build an integrated network supporting bandwidth of 50 ~ 100 Mbps to 20 million subscribers via wired or wireless access. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA * In January 2005, Northrop Grumman Mbps DSL line might have at most 200 convey to Bill St.Arnaud. We have been awarded a multi-year contract to Cas- flows with a resulting variance gain of trying to add traffic types like voice pian Networks for joint development of a 14:1, still 4 times less variance than 16 and video that require call acceptance critical element of the space communica- queues. But given call acceptance control control to a router designed mainly for tions system payload for the Air Force’s for voice and video it is quite reasonable TCP traffic and the same control method Transformational Satellite Communica- to achieve very low delay variance and (WRED) that works for bulk data does tions System (TSAT) Space Segment. low loss on these UDP flows if no new not work well for streams. flows are accepted once the average UDP Coluccio: (To Larry Roberts) I had lost load exceeds some level. Without call ac- [Editor's Note: I am continuing to wield site of your involvement and pursuits ceptance control, there is no way to keep a machete chopping chunks of sections within Anagran, the company you found- voice or video from overloading the link and and entire section on IMS in the Moed and currently head up (I believe), and all the flows having significant loss. bile World -except that is, for Sebastian whose mission is to: The TCP traffic can fill the rest and may Hassinger's closing remark that follows.] have higher loss than in the core, but it “ ... [develop] a next generation Customer is the streaming flows that need to be IMS in the Mobile World Premises Equipment (“CPE”) IP Router. protected. This works well at the edge Anagran will enable high quality voice and of course this is where it is needed Sebastian Hassinger: IMS seems to me and video, load balance among users, most. One can improve utilization and to be a layering on top of the architecachieve high throughput TCP especially reduce cost in the core as well, but this is ture that the mobile network operators critical over long distances and allow not strongly related to QoS, more the fact have been developing for the delivery of fairness of broadband utilization.” that FSA routing is less expensive than network-delivered value-add services to packet routing and can achieve higher their customers. Last year this was called http://www.anagran.com/index.html efficiency. Service Delivery Platform, or SDP. From I’m therefore particularly eager to hear what you have to say about how I characterized the edge/distribution/access traffic handling considerations for residential and soho triple play traffic in the message immediately inline, quoted above. Roberts: Frank - As you have realized in reading the Caspian announcements and my Anagran site, I was talking about individual flows, not MPLS composites. If we then consider a Flow State Aware (FSA) router that can identify all flows in an IP link and manage the QoS (discarding, scheduling) of each flow separately, rather than managing a small number of queues, then the statistical gain is far improved even at the edge of the network where there may be much smaller links. IP generally has 100,000 individual flows per Gbps of traffic. In the core then one has 1M flows per 10 Gbps link and by standard queuing theory if one manages all million separately rather than as 16 queues, the improvement in variance is 1M^(1/2) or 1000:1 rather than 4:1. At the edge, the gain is less, since a 2 As to OPEX, the current methods of managing QoS in standard packet routers with a number of WRR priority queues does take considerable manpower to tune the limits on each queue as the traffic mix changes. This tuning need not continue however once the routers are measuring and controlling all the flows and automatically controlling the accepted load to insure the QoS of all accepted traffic. There should be no need to tune or monitor the network manually at such a fine level. The QoS is being managed automatically and the network operator only sets up the global parameters once and then assigns each user one of a number of standard profiles. The network then enforces these profiles to maintain the global parameters. With the current “best effort” router design it is costly in both OPEX and CAPEX to improve QoS and there is no way to control loss if too much video is requested. But this situation is a result of the 30 year old router design, and we should not assume it cannot be changed. This is the message I was attempting to 128 a bird’s eye view each new TLA (Three Letter Acronym) actually looks like another incremental step towards the same eventual goal. Namely, leveraging emerging telcom and IT technologies to build those fabled Next Gen Networks (NGNs, the TLA from the year before last) that allow network operators to fulfill their dream of becoming Service Providers of value-add services to their captured customers. IMS specifically is the turn of the crank that adds SIP on top of an end-to-end IP layer that promises, like every turn of the crank, to lower operating costs while enabling the efficient delivery of new revenue-generating services. In some conversations it seems that IMS is primarily preoccupied with the delivery of video to handsets. It also incorporates some principles of SOA (Services-Oriented Architecture) and Web Services. In other words it’s mostly “marketecture” wrapped around the integration of a few new bits of technology into the existing network. Symposium Discussion March 3 -10, 2005 Some VoIP Regulatory Issues Highlights Vonage Whining its Way to Open Access Editor’s Note: Early March saw an example of what will undoubtedly happen again: port blocking used by a small North Carolina CLEC. Vonage asked the FCC to step in and prevent the blocking. On March 3 the FCC ruled in Vonage’s favor. http://www.internetweek.com/allStories/ showArticle.jhtml?articleID=60405234 “The Federal Communications Commission announced Thursday that it had reached a $15,000 consent decree with Madison River Communications, a Mebane, North Carolina service provider that calls itself the “17th largest phone company in the US,” with “234,204 connections in service.” According to the FCC, Madison River pledged to “refrain from blocking VOIP traffic and ensure that such blocking will not recur.” This touched off a lot of discussion. I am publishing only a small part of the discussion because it goes too deep into the American regulatory arcanae that in the view of some been responsible for the failure of broadband infrastructure in the US to keep up with the rest of the world. Matt Wenger: Vonage’s response is perfectly indicative of what is wrong in Telecom. The infrastructure providers want to have a monopoly on services (we own the customer.) The pure service plays want no-cost access to infrastructure (we own the customer and you, Mr infrastructure provider, own the duncecap) The customer wants a wide variety of affordable services with as little hassle as possible. This fleeting moment where the rules of engagement have fallen behind the ability of the technology (read: loop-holes) is not sustainable. It is acceptable for Vonage et al to take advantage of it while it “fleets”. It is childish to whine about it when it “flows”. We need a sustainable model and what we’ve got now, Mr. Vonage, is not it. If you ain’t part of the solution... Goldman: Of course, many CLECs have argued to regulators that competition demanded open access. Some VoIP providers are former CLECs who now say the CLEC model was never valid. One CLEC notes, “The worm may have turned over a bit.” Coluccio: If the ILECs knew how to deal with ‘channel partners,’ then the CLECs would have been regarded as distributors in the supply chain rather than being viewed as adversaries. Being the bigger of the two segments, the ILECs could surely have set the stage as such. “Some VoIP providers are former CLECs who now say the CLEC model was never valid.” Why is it that when a CLEC upgrades to VoIP its name is changed to “VoIP provider” or an “ITSP”. And when an ILEC does the same thing, i.e., upgrades to VoIP, they are still called an ILEC? Editor’s Note: While I have eliminated many exchanges from the mail list discussion -- On March 3 Matt Wenger again sums up: Those that insinuated against my insinuation (Dave, Henning et al) basically fall into the same camp: That the customer pays their ISP, Vonage pays their ISP and so everyone pays their fair share. That was not the point. The question is, is this model sustainable? The answer is likely not. What is the pressure on local loop Internet charges to the end user? Is it trending up, keeping pace with inflation? Or is it trending down? What’s happening to the costs of the Infrastructure provider? Those pole contact fees going down? How about OSP maintenance? Is the industry of Internet infrastructure becoming more like a diverse group of Small specialty Organic farms or trending more to Mega-agriculture? As internet con129 nectivity costs approach zero (supply continuing to outstrip demand - chapter 11 instead of chapter 7 etc...) where do the revenues come from? Triple play? Quadruple play? The be-everything-toeverybody-play? How does the LLIP protect their investment and their revenues? Is this unfair? What happens to Vonage if all the local loops go out of business? (10 bucks says they buy them - wait a few years for those announcements to start happening - control is ultimately in the pipe). So I ask the folks who didn’t like my insinuations, if I own the pipe, why can I not block or otherwise impede traffic that is counter to my business model? If I want to sell VoIP, why can’t I block yours? AOL has been redirecting web sites for years (or so someone once told me - though I personally saw no proof). Can I block porn? Can I block things that might get me sued? What if I only open certain ports? Technically, I’m not blocking. I’m just selectively allowing. What about destination-based policy routing? Are you going to come after me for that? Are you going to force me, via whining to government regulators, to offer great QoS for your packets?? What if my QoS isn’t as good as the next guys and I cause you a bunch of expensive support calls? Are you going to force me to have the latest and greatest kit all the time and an oversubscription rate of no more than... (hey... I’m liking the sound of this for us gear pushers).... hmmm starting to sound like an SLA... If you want to control the medium, buy it. Or rent it. Offer cash incentives (ok bribes) back to the network providers. Do joint marketing (wholesale/retail). Hell do anything, but for Pete’s sake (and mine and everyone else’s) don’t go whining to the government of all places to rere-re-regulate this god-forsaken industry (Francois, I loved “re-regulation”, although perhaps more accurately it would have be re-re-re-re-re-regulation. I will COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA steal it and use it re-re-repeatedly!). Amazon pays for bandwidth - for website traffic that is probably an acceptable metric for now (I haven’t seen another compelling model, although ad insertion is really starting to gum up my models at the moment... more on that another time). As Andrew and others have demonstrated so well on this list, and I continue to re-re-reiterate - look at cell phones. You must move past charging for commodity bandwidth and into charging for VALUE if you are to build a sustainable network model. Just because one service warrants a mbps fee, doesn’t mean the next does. We’ll never get HD Video over IP (which means the triple-play will be re-re-reclusive which means sustainability issues for small infrastructure providers). Mr Whine E. Vonage, blocking or otherwise playing with packets bound for your switch impedes on neither your right to freedom of speech nor ours. You have the right to build your own infrastructure. Or, W.E. Vonage, you could do what the rest of us who don’t like system are doing, and contribute positively to changing it by supporting open models where you can control the quality of your service. Pay your way or pave your way to the center of the earth with the others you blind bastard. Your “call” (to coin a phrase... ha-ha! the re-re-entrendres are endless! ) Schulzrinne: Since access providers have a near monopoly in many communities (and are trying very hard to maintain such monopoly - see community wireless discussion), they are not just another business that can choose what service they offer to you. Wenger: Under what statute is a CLEC forced to provide ANY service? The only folks I know who have to provide service are the incumbent LECs and all they have to provide is Phone. No one has to provide video, voice over IP, internet access. If someone owns the infrastructure they can choose to sell you whatever they want. Furthermore, they can sell you what you don’t want. And they can refuse to sell you what you actually do want. Unless the law changed while I was asleep, this is the sit-comedic nature of American Telecom. Schulzrinne: You can’t claim the benefits of monopoly and common carrier status and then complain when you can’t discriminate on services. Wenger: Wha-ha? Sure you can. You can do anything you want in a deregulated environment where you own the asset. Watch what happens when these guys build fiber. Have you seen the FCC ruling? Now you are right, we are in a weird grey space of non-re-un-re-regulated telcom, but on a basic principle if I own the asset, and I am a private company, I can do what I want. This is called de-regulation and it is where we are going, like it or not. But let’s drop the ILEC case - I’d love to bully the ILEC as much as the next beanpole with a penchant for regulation/public floggings. What if the company that Vonage is complaining about is actually a CLEC? Would this change your argument? Would you say, oh well no, I guess they can block? What if it was the cable company (which my badly informed, usually delusional contacts tell me this is the case)? What if it was the local wireless ISP? What if it was all three and they were all competing in the same geographic area? Are you arguing that no company with an asset should be able to control that asset (and protect their investment) or are you just arguing that the assortment of ILECs can’t have that right? Schulzrinne: This illustrates the need to separate out the rate-of-return, natural monopoly parts (fiber, coax, copper, licensed spectrum) from the competitive aspects of service provision. Wenger: Now you’re talking my language. I couldn’t have said it better myself. You would have to look a ways to find a greater believer in (or advocate for) pure Open Access than I. And you’re right. That’s exactly what this illustrates. But when you do that, how do cover the costs of the Infrastructure? I won’t rehash my arguments from November/ December here. The point is that simply separating the services from the infra130 structure does not sustainability create. What is the billing model? If you say mbps then I won’t be investing in your network anytime soon... Like it or not, Data is not Data is not Data. Just like steel is not steel is not steel. Some steel is a knife. Other steel is a car. Other steel is the plate in my head (where is that medication anyway...). (even better is water is not water... I’ll pay more for water in a bottle in a store than from my tap poured into an empty bottle. I’ll pay even more for that bottle on hot golf course.) Basically Henning, I think we hold the same belief. The ideal would be separation of infrastructure from services. What I think we disagree on is how the dollars might flow to sustain such an ideal. I’d love to hear a different model of sustainability for an open system if you have one. While we’ve had a lot of success (after a lot of failure) with our models, there is always room for improvement or even radical innovation. Schulzrinne: There is no natural right for a company to leverage public rights of way to extract maximum monopoly rents. Wenger: Err.... I think I’ll drive by a debate on natural rights. (But on second thought) I have no natural rights to land my house sits on either. Somehow though I have ended up with for-all-intents-andpurposes-natural rights. The American Dream, I believe it is called. Property law. But you are on to something here. BOY WERE WE DUMB HEY? They asked and we gave. Sometimes, in our slightly less drunken moments, we even thought to sell. Schulzrinne: Nobody forces Verizon or the cable company to be in this business. If they’d rather sell that part to a true utility company, I’m sure there would be takers. You are essentially advocating to go back to the French minitel model; there were good reasons it didn’t survive the arrival of the Internet. Wenger: Yes and none of them had to do with the billing model. Support, new ap- The COOK Report on Internet May June 2005 CODECs and Perceptions of Voice Quality Coluccio: On the matter of voice quality, most assessments are usually vague and subjective, such as “pretty good,” “toll quality,” or compared to what were one time, and continue to be in many cases, similarly inferior gradings such as “cellular-like” and “citizens band-like” quality. Is there a published reference that someone here can point to that charts the frequency response ranges and other applicable parameters affecting sound quality of all codecs used in VoIP today, especially those used for Skype and the more popular variants of SIP, H.323, etc. for comparison purposes, even if those parameters only reflect ideal conditions? The only time I’ve experienced SIP to be as high in quality as Skype has been when a properly engineered enterprise LAN was used, and less so on properly designed intranets. In either case, this has been true only when there were no intervening switched services involved, such as those that would employ an H.323 gateway to the PSTN. Those, as a matter of physics, default to the narrower bandwidths found in PSTNdesigned multiplexed channels, i.e., they never allow more than 3400 HZ of passage, and sometimes that number can be 2700 HZ. In this message I use the term “artifact” consistent with Webster’s fourth definition: an object, observation, phenomenon, or result arising from hidden or unexpected causes extraneous to the subject of a study, and therefore spurious and having potential to lead one to an erroneous conclusion, or to invalidate the study. In experimental science, artifacts may arise due to inadvertent contamination of equipment, faulty experimental design or faulty analysis, or unexpected effects of agencies not known to affect the system under study. Having stated that, I’ve encountered more forms of auditory artifacts on SIP sessions on intranets than I have with Skype on the open ‘Net, even when the majority of those conversations have been near-pristine for the greater portions of those calls. The artifacts result from, I have to surmise, spikes in neighboring applications. Skpye, too, has resulted in momentary noticeable blips and dropouts, but far more seldom than my SIP intranet experiences, thus far. And neither has been able to quite match SIP in the latter regard when a properly engineered LAN was the sole medium used. But I have been dropped stone dead (off-hooked) from several Skype calls over the past month for no apparent reason, and with no means of determining why those disconnects occurred, which, I believe, may be one of Skype’s Achilles heels where enterprises are concerned. Skype simply doesn’t permit the visibility into the network, as things stand today, that is needed to satisfy corporate surveillance and OAM requirements, even if some of those requirements are, in the final analysis, perfunctory, at best. Again, a summary reference for codec designs would be useful, and if anyone can point me to one, or explain why such a tool would be less than fully useful, I’d appreciate it. Reed: Voice quality is mostly about psychoacoustics, not frequency response. I have lots of references in my drawer about that. Unfortunately, there isn’t a way to analyze waveform distortions to determine “voice quality” without user testing on a variety of voices. Gill: You can use MOS-CQ scores to get a pretty good idea of how it works. Brix makes boxes that do this for you. Reed: Generally, however, end-to-end codecs work a lot better than intermediate translations between codecs, given a particular bottleneck bit 131 rate. This is why Skype can work well - it never attempts to use existing voice paths in the middle. End-to-end SIP can be excellent as well. But too many VoIP providers make their business around interconnect, and then do the interconnect poorly. It’s well known that certain digital cellphone codecs interoperate poorly, even when operated alone they provide quite good quality. I don’t know of anyone who has worked on “impedance matching” (in the metaphorical sense) between low-bitrate psychoacoustic channels that work on slightly different theories - that’s what it would take. Gill: Is there a reference you can share. I regularly talk between CDMA (using EVRC 8k and 13k et al) and GSM forcing AMR or EFR EFR/HR, and at least to my ear, the background noise and poor noise suppression of the GSM codecs tends to dominate the call quality issues. Reed: Vijay - I don’t have a reference for composite-codec distortions, just comments from many peers in the cellular systems engineering field (Qualcomm, Motorola, Ericsson, Nokia) about the problem. I’d be interested in a reference as well. It’s plausible, given the kinds of tricks used in different codecs, that artifacts that are not salient to a human ear are detected and distorted by a second similar or dissimilar codec into distressingly audible artifacts. That certainly happens in image and video compression (for example the edges of blocks in JPEG are not visible, but if you then take a previously compressed JPEG image and run it through certain wavelet compressors, the edges are emphasized and become an irritating grid like artifact. Such artifacts are not visible/audible in simple measures like the frequency response, but are nonetheless highly visible/audible to the nonlinear detectors we humans have to work with. Hence the benefit of using a single end-to-end compression codec, rather than concatenated codecs willy nilly trying to simulate POTS. Coluccio: David, your points concerning psychoacoustic considerations are well taken, although I was not looking at the matter from a psychoacoustic-, or even psycho physiological- standpoint. Whatever the deeper analysis may provide from those standpoints, however, one cannot discount the heavy weighting that frequency response has in determining what we generally perceive as voice quality, especially under *ideal* conditions. And it was the ideal condition that I was asking for - not tandem, dissimilar, connections in the presence of congestion and other forms of noise (latency, jitter, “impedance” mismatching). By the way, I’m glad for a change to see someone recognize that the latter term ‘impedance’ is in fact a form of metaphor when used in this context, more often than not ;-) I’ve not seen actual studies conducted on the mixing of tandem sections of several VoIP paths, i.e., “in line” sections connected to one another using multiple forms of coding. But I have seen in the past TDM equivalents, where, for example, proprietary forms of digital speech interpolation (DSI) and delta modulation techniques of the type that Tymplex used were connected in back to back configurations with ADPCM, and then with cellular .72x algorithms, with results that were sometimes utterly horrific. For the hell of it, I’d still like to see the chart I asked about earlier, if anyone has one. I’ve Googled this thing to death without reaching any satisfaction. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA plication development, integration, opex why Powell’s policy is so, frankly, cor- choice. Thus I suggest that Powell is posbut not the revenue model: rupt. The ILECs have monopoly power, turing, using Vonage’s temporary, limited which has been greatly strengthened by competition as a bad excuse for giving From Wired 2001: “the real story is recent FCC actions. (They fought the the ILECs almost all of the cookies. Minitel’s ability to convert online traf- USTA and USTA II cases about as hard fic into money in the till -- not only for as Bre’r Rabbit fought being thrown Pershing: This is distinguished from, France Telecom but also for the 7,000 or into the briar patch.) The Telecom Act say, I don’t know, SBC, and its market so companies that depend on the bulk of laid out a fairly clear vision of competi- dominance of the infrastructure facility. their business from teletext customers. tion among certificated LECs, providing It is questionable whether the infrastrucFrance Telecom raked in over 690 mil- common carriage to unregulated ISPs. ture market can be competitive. But lion euros ($614 million) in revenues Powell does intense violence to that perhaps the application service provider from Minitel services in 2000, 440 mil- law, making life even more difficult for market can be. lion ($392 million) of which was then facilities-based CLECs (I’m *not* talkdivvied out among participating com- ing about UNE-P here), and proposing Goldstein: That was the Telecom Act panies. These revenues, generated by the removal of common carriage protec- model -- unbundle the parts of the netcharging users between 0.6 euros (50 tions needed by independent ISPs. So work that were needed for competition, cents) and 1.35 ($1.20) euros per minute, the ILECs and cablecos will have 100% or whose absence would impair competidepending on the service accessed, are of the cookies in the jar to split between tion. Nobody but nobody though, when it was passed, that ILECs would be able collected as part of customers’ bimonthly themselves. to proclaim that there was no need to unphone bills.... <snip> But analysts still give Minitel a two-to-five-year window Since the cablecos are, in most cases, bundle the local loop, or for that matter late to the voice party (they’ll create transport to the CO where the loop ends! for survival. a strong second provider within a few Yet that is precisely what the TRO all so That’s enough time for France Telecom years, and are starting some aggressive often does. and its partner companies to continue PacketCable rollouts already, but few making healthy profits from Minitel - people have seen it yet), the ILECs in Sensible public policy is to split the ILECs - and e-commerce firms to try to figure early 2005 now have almost all of the into outside plant companies (LoopCos) out how to work the same magic for the voice cookies. UNE-P was the primary and service companies. CLECs and ISPs Internet.” mode of residential voice competition, could then compete with the latter, using and its phase-out is so fast that most whatever technology floated their boat. Check out Andrew’s analysis of the cell players don’t have time to evaluate, let phone service-bandwidth costs table. A alone install, UNE-L replacement (which Franscois Menard: Hear hear, Fred has call costs “less”(in $/mbps) than an email as noted has gotten a lot harder since the summed up what’s wrong in the USA like which in turn costs less than an SMS. Remand TRO). So there is no real voice I have not often seen in so few words. Data is data is data right?? Wrong. What competition *except* for parasitic VoIP This is a keeper for posterity and likely to is a shame is that they didn’t figure out operators like Vonage. be circulated to a few CRTC staffers who how to work the Minitel magic “forward” are wondering what’s so wrong with the into the Internet. (<start plug>Although I These guys are no threat to the ILECs. In FCC the next time they ask me. know of a little Swedish company that order to use them, you first need DSL or did... :-)</end plug>) a cable modem. In the case of DSL, you Editor: Earlier on our network Economneed a first line from the ILEC before ics list Francois commented: This [the Editor Again: Finally On the Cyber you can get it. So Vonage becomes a FCC order by which Madison River Telecom Law mail list on March 10 Sean cheap LD play, and a privileged offerer agreed not to block Vonage and paid Donelan Asked: Why is the FCC wanting of Foreign Exchange service (which non- a $15,000 fine for doing so] is a nonto act on Vonage’s behalf? VoIP players can’t offer except as costly binding staff recommendation associated LD Access). In the case of cablecos, with a consent decree, which to me is Genny Pershing wrote: “Competition.” Vonage can ride along for now. But once far from a commission decision with full PacketCable is available from the ca- legal force. Until next time. The traditional primary purpose of com- bleco, Vonage, with its necessarily lower munication is competition, or lack there quality, won’t be such a great deal. Ca- Coluccio: I don’t know about that Franof. It is not that Vonage uses VoIP. Do bleco costs of offering PacketCable are cois. Alternatively, a provider may simply not superficially confuse policy with the low enough (self-parasitic) to kill them choose to de-prioritize voice by devising a means of granting higher status to evmeans by which policy is applied. VoIP off, if they deem it necessary. erything else that is best effort, if they brings competition. Therefore VoIP is the magic line drawn in the sand that distin- Since there appears to be no legal basis wanted to be real scum bags about it. guishes policy. on which to ban VoIP port blocking or filtering by cable ISPs, Vonage as a com- No, the way that the incumbents will Fred Goldstein answered: And this is petitive alternative to the ILEC is a shaky win share is by prioritizing their own 132 The COOK Report on Internet May June 2005 time-sensitive services through DOCSIS/PACKETCABLE and VoDSL/ATM QoS mechanisms, while best effort continues to ride what’s left over. And the telcos/cableops will charge an additional ten to fifteen bucks per month more than the parasitics while they’re doing it, in my opinion. I don’t think that any big names want to be associated with the alternatives having to do with blocking. But you never know. How to Explain Why Skype Works Better? Retzer: How do you explain that Skype simply seems to work better than H.323 and SIP? Neither of those would, I think have met the Sherpa test as well or the link reported yesterday WIMAX local - - Internet - PSTN - T-mobile even assuming they found their way to the PSTN. The quality probably would have been poor. I gave up on Vonage locally last year because quality was so poor over my Comcast account. Skype works great, at least on the limited tests I’ve tried. I’ve similarly found h.323 calls need to be engineered to work well but Skype , as they say “just works.” Not being able to look into how Skype works or how the flows actually go (which is why I asked if you or anyone else has tried to analyze flows), one hypothesis I have is that is trading latency for jitter and bit loss - accepting delays to otherwise improve the quality, maybe using P2P to cache calls. Obviously, codec choices may also have a huge bearing. Your comments, below seem to support this notion to some degree. Have you done any actual analysis of Skype? Thanks Forster: Better execution. Architecture counts for about 10% of a solution; execution & operations count for about 90%. In this case the area of great execution is that Skype apparently downshifts to lower bandwidth and/or larger packet sizes quite nicely by itself. There’s no question that the codec in the phone Cullen sent could do a decent job on a 64kbps link, but neither Cullen nor I had the time to fiddle around with the parameters. As Cullen mentioned earlier, Skype licensed a codec and packetization engine from some outfit and they’ve done a great job with it. I’m happy to give Kudos to Skype for their very nice job, but let’s not confuse a great product with other architectural and protocol issues. Skype won’t be the last really great voice software product. Reed: Hear, hear! Jim Forster - you nailed it with that line. But the folks who think they can sit this out in comfortable enterprise niche markets, selling FUD to the CIO’s about requirements for perfect QoS, -- in order to win closed, proprietary, incompatible bids that interoperate merely as crappy interconnects with the PSTN -- those guys are toast in the long run. Remember that Cullinane (king of the database market) *deigned* to do a crippled PC version of their software that was carefully designed not to cannibalize the customers for their main product. Where are they now? Wall St. “loved” their embrace of the PC market, by the way, and thought they were really “cutting edge”. Shows you how smart it is to listen to the investment bankers on these things. Everybody loves to suck up to the selfdefined brilliant CEOs who “bless” the new technology with their charismatic narcissism. (Leaders are narcissistic by definition, which is what gives them charisma, often a very good thing in a leader, so that’s not intended as a criticism; rather I am emphasizing why it leads them to be somewhat blind and insulated by sycophants from really disruptive, worrisome changes) I doubt that Skype thinks it has won the market. I think they are still scared, which is a much better attitude than those who think SIP is merely an incremental way to do the same old stuff over IP. I can’t emphasize enough that there *is no enterprise market* for communications that is separable from the dynamics of the broad telephony market, the weak separation only being a sales channel separation and no more. This was Lotus’s big mistake, and why Notes didn’t own the Internet when Notes could have become the “better WWW”. As I was 133 leaving Lotus in 1992, I was one of a few people that kept telling my fellow VPs that the Internet was either a threat or an opportunity, and they had to get out of the enterprise-centric sales model, starting to look at making Notes architecture just work as a platform across the entire inter-enterprise and inter-SMB and interperson space. They chose margins over market share, and came too late and too incompatibly to the Internet (with Domino). Again the analysts “loved” Lotus’s embrace of the Internet, and believed John Landry (remember Cullinane - he was the guy who conned the analysts at Cullinane) that Domino was going to define the real enterprise Intranet. I know that these stories of mine should scare the CEOs, but their cockiness, and that of the VPs of Sales who “own” the accounts, will ensure that they don’t hear what I’m saying. Remember that *sales* sees the enterprise market as distinct, and will invent reasons to believe that despite evidence to the contrary. (And most CEOs come from Sales...) So, the real hope that SIP will have a chance to be the defining architecture lies with true entrepreneurs who hear what Jim says above, and realize that SIP is best deployed by someone who does not see it as an extension of their business, but as their ONLY business, end-toend. They won’t exclude the enterprise, but they won’t allow the enterprise to “own” the evolution of SIP either. Who will be the Cisco (who played that role with IP in the early days of the Internet) of this space? Cisco is not the Cisco it was - it’s now run by the charmers who make Wall St. happy - the Sales guy. Pulver has a shot, but honestly, he makes too much money on VON to be scared enough to make it happen. Cringly Asks Have Best Days of VoIP Come and Gone? Mike Cheponis: Cringley’s column presents an interesting take on what we’ve been discussing here. http://www.pbs. org/cringely/pulpit/pulpit20050303.html COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA The Best days of Voice-over-IP Telephone Service May Already Have Passed By Robert X. Cringely These are heady days for Voice-over-IP (VoIP) phone services. From Vonage to Packet8 to Skype and a hundred more besides, several million people around the world are enjoying really cheap phone calls that are carried primarily over the Internet. But that fun may be diminishing soon because the big Internet service providers, which is to say the big telephone and cable TV companies, are about to start taking back that third-party VoIP traffic, leaving Vonage and the others at a distinct disadvantage. [Snip] The other big broadband Internet provider is typically your local cable TV company. Not sharing the values of phone companies, they ignored VoIP, too, until they realized that it could become a new source of revenue. With so many things available for free on the net, finding a service people are willing to actually pay for is thrilling to companies like Comcast. Now they just want to make sure that the VoIP service you use is THEIR VoIP service. The trick for phone companies and cable companies alike is to hurt the VoIP upstarts without incurring the wrath of Congress, the FCC, or any other regulator. They have to be sneaky. rules for passage and a private highway lane to drive on. The net effect is that any packet that isn’t tagged will only get “best effort” service, which means whatever is left. “Best effort,” as defined by IETF RFC 791, makes almost no guarantees. The packet may arrive damaged, it may be out of order (compared to other packets sent between the same hosts), it may be duplicated, or it may be dropped entirely. And that was in the good old days. Now imagine “best effort” transport on a backbone that is already clogged with tagged traffic that gets preferential treatment. Where previously all packets got “best effort,” in this new system some packets get better than best effort, which means the remaining packets will effectively get worse than best effort. The telco and cable guys know enough about their networks that they can throttle their network capacities up and down so that “best effort” service is going to be pretty awful. But have the magic tags on your packets and you’ll have decent service. The beauty of this approach is that they’re NOT explicitly doing anything to the 3rd party service applications. They’re just identifying and tagging their own services, which is within their rights. [Snip the remainder] Coluccio: Thanks, thanks. Cringely needs to read the COOK Report:) Here’s how they plan to cripple the Vonages and ‘s, according to friends of mine who have spent 20+ years in engineering positions at telephone companies, cable companies and internet service providers. As the phone and cable companies begin offering their own VoIP services in real volume, they plan to “tag” their own VoIP packets so that at least within their own networks, their VoIP service will have COS (Class of Service) assignments with their routers, switches, etc. They also plan on implementing distinct Virtual Local Area Networks (VLANs) for the tagged packets. I don’t want to listen to any Volkswagon ads. Nor do I believe necessarily that he is even headed in the right direction, but a lot remains to be seen concerning the sustainability of the cable and DSL models, all migrating to triple play, as it were, and what the breakdown in true unaffiliated broadband will look like over the next several years, when true condo-like Layer 0/1’s are deployed. Tagged packets get both less restrictive My belief is that video technologies Have you or anyone else here attempted the switchboard.ca applet? 134 needed to satisfy residential entertainment requirements are slowly breaking through the barriers by the established by the lock-in frameworks of the established cable industry. Now we see the RBOCs emulating this model to a great degree, where they are not imitating it outright, despite their purported plans to use IPTV as a foundation for content delivery. But how long will it be before these established models start to take on anachronistic attributes, especially where users have ample bandwidth to try out emerging alternatives? In contrast to having users fed the standard pablum that service providers have regimented themselves around, users are now seeing that the light that shines at the end of the tunnel is illuminating a new kind of video - and other forms of content - acquisition by executing downloads directly from Web-based content providers of their own choice. For this proposition to be lead to a sustainable model, most of the so-called (bandwidth starved at this time) “broadband” FTTP or HFC or BWA access links would have to be freed of their usual lists of “must also get” payloads that ride over them today, which are, in the main, represented by vertically integrated services over ATM/DSL and FDM/DOCSIS, respectively. In other words, give me a pure multiten-megabit or gigabit pipe to my home and I’ll select what I want, when I want it, and talk to my girlfriend in Bulgaria using any service provider or PC-enabled application that I choose. It could even be the PSTN, for that matter, but I’ve made the choice at that point, and not my State capital’s favored son. Cringely cites: “The phone companies -- always slow to react -- did not pay much attention at first to these outfits they collectively refer to as “parasites,” but now, with several million lost customers, they are paying VERY close attention indeed.” Silly me. I’ve been using the term “parasitic VoIP” as though it were an accepted element in our shared, albeit evolving, The COOK Report on Internet May June 2005 lexicon. But I think it’s clear that Cringely is suggesting that the term ‘parasite’ is being used pejoratively. Any comments on this, before I further insult someone inadvertently through its continued use? Menard; Yes, but he neglects to even mention that whatever evil play may be in sight of ILECs and Cableco’s. Such plans would fall flat faced if the rules of the game are applied diligently which would mean that prioritized virtual local loops (i.e. packetcable loops, AAL2VoIP VoIPoDSL or VoIPoVLANoEFM(Ethernet in the first mile)) diligently and properly unbundled at essential facility rates (incremental costs + 15% mark-up). This is what we’re working on in Canada (with zillions of words on this topic on the public record already) and I think the CRTC has no choice to deliver and WILL ACTUALLY DELIVER, including this over DRY COPPER as it is currently the case over Coax in Canada. Why it is not happening in the US is purely a failure of the FCC to be proactive. Now let’s see if the IP-enabled services NPRM motivates the FCC to do something this time? Coluccio: Francois- You’ve asked many times why the FCC hasn’t done as much in the USA as the CRTC has in Canada. To do what? To perform surgery on the rules governing a dying model? There was a time, maybe five, six years ago, at about the time that AT&T picked up the assets of TCI, when I was at your side seeking equal access to DSL, DSOs and best effort capacity over cable modem links. Recall all the discussion over where the IP access point would be situated on the cable modem model? I’ve watched you hone your regulatory acumen since those earlier times, to the point that I’ve enjoyed reading some of your poignantly written petitions to both regulatory agencies. But it occurs to me at this later date that you are continuing to fight a battle, which, if successful, may be only a Pyhrric victory, at best. Just so that what I’ve stated above is not entirely misunderstood, certainly there’s life left in the old model, and certainly some benefits will accrue if incumbents are forced to open their capacity to competitors. But I submit that your energies could be better spent working towards the creation of a new model, instead of seeking fairness treatment on the old one. In all fairness to you, however, you’ve also done your fair share in seeking new approaches to last mile delivery, and you’ve done this quite capably in your work in the areas of fiber to the home and condominium cable architectures, both in the outside plant and the inside (neutral) hubs. All of which makes it appear even more curious to me why you’ve split your time the way you have, doing battle on both sides of the divide, i.e., between legacy and nascency. Cringely’s view, by the way, merely reiterated what I stated earlier to you, yesterday, concerning the other ways that the incumbents have to skin a cat. And that was, namely, by relegating competitive packets of all types to the bottom of the best effort heap. By the way, the 15% on top of marginal costs to provide incremental or shared capacity is what a provider needs to merely subsist, if not even a little bit more. This doesn’t say much about why a service provider might want to even stay in business. But I’m open to correction on that one, as I am to all of the above, as well. Thanks for your reply. Menard: Frank, If your message had come from anyone else, I would have said that your message had been written by someone who doesn’t understand the challenges of competitive entry. example: twice in my life the US Postal Service tried to get Congress to block “electronic mail” because they were supposed to have a government monopoly (including being the largest jobs program in the United States!) according to the law. Similarly, twice the US DoD tried to get all civilian use of strong crypto banned under ITAR regulations, US patent law (which protects munitions, which include crypto). These were serious efforts, involving Congress and the various agencies. They failed to stop both revolutions. Smaller versions happened when CATV and Satellite dishes started. So you can be forgiven because you are young, and revolutions are necessarily rare. (You don’t want a state of continuous revolution, believe me). But don’t let the regulatory process suck you into the idea that it is the place to effect change. Menard: Given this restriction, and post bubbe, I have come to conclude after seeing other (much better VC’ed) competitors fail that we must finance our entry from proceeds generated from the value-added resale of unbundled elements under the protection of price floors. You will notice that our condo fiber entry has also been accompanied by several price floor regulatory enforcement exercises, because if you’re an ILEC and you’re dominant, you can always bid below cost to kill a competitor then afterwards raise prices to defray these costs. Why, because ILECs can, and infact HAVE been doing this in Quebec as a result of a mere 150 M$ subsidy from the Quebec government after being unsuccessful in lobbying out the program. Reed: Ever the optimist, I just refuse to cede such power to the regulators. Of course that’s what the *tendency* is, Francois. So our model of condo fibre on the backbone and WiMax in the access, faces similarly the wrath of forborne-conditional-to-the-presence-of-cost-basedUNEs ILEC DSL where forcing out the enforcement of price floors is much more complex and intertwined with the provision of local telephony (be it TDM or packet-based). But the regulators can be beaten. Just for So just to make sure we do not eat our Evolution is the only way in telecom because regulators will not let revolution happen. There are too many jobs at stake. 135 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA shirts deploying WiMax, I must fight it out on the DSL and DOCSIS fronts. If I cannot finance WiMax because of the risks, then FTTH is out of my league, at lest for the next two years. COOK Report: I think what Francois is saying is that the Canadian incumbents are now under bidding Xitelecom (his employer) for condo fiber builds? What I didn’t quite understand was the incumbents have been able to do this courtesy of a $150 SUBSIDY from the Quebec government. How in the heck did they bring THAT about? Is there some Canadian equivalent to Verizon in Philly underway? Francois or someone - tell us more please. I think BOTH Frank AND Francois have compelling arguments here. But Are P2P Voice Applications Blockable in the Same Way as Vonage or Lingo? Meanwhile I have a technical question. we have seen it mentioned that the cable plant can use docsis to descriminate against Vonage, Lingo and the like. Would really appreciate it if one of our more technically minded folk would comment on the dangers here for VoIP traffic. What is the common denominator here of susceptibility if anything? SIP? Something else? Port assignment? The FCC doesn’t like port discrimination but unless Brand x is upheld it will make no different because the cable cos are regulation free. (Without going back to the article) was he saying that DSL telco folk could do the same thing? And finally and most important what about the P2P services like ? How susceptible are they to traffic shaping and prioritization services? Has anyone tried to strangle a p2p service yet? They have gone after them legally but could the incumbents kill in the last mile? I would have thought not but i certainly don’t know the technical detail the way many of you do. Reed: You can *try* to block , Vonage, Lingo, Bittorrent, etc. at the DOCSIS layer, but the fundamental issue is “recognition”. Recognizing a specific end-to-end application can be done by looking for “signatures” specific to that application. To do so, you have to reverse-engineer the application, and distinguish it from other applications you dare not block (because they would cause customers to switch to another provider, at least in a competitive market). Thus, I could probably come up with a scheme that blocks Skype, but if I and many other providers start blocking enough to annoy Skype, it can release a new version that doesn’t have that signature. Ultimately this “arms race” can be won by the ISP only if a) it can keep its customers from switching to a competitor that does not block (and in the US, oligopoly attempts to organize competitors violates antitrust laws, so you can’t do this by forming a club of thieves to conspire to define a blocking strategy across all competitors in a market), or b) it can find a class of behavior that it can block that is acknowledged to be “bad” by all competitors and the government. Linking the blocking to that class is possible. For example, one could enforce a rule that every 10 seconds the access port refuses to send ANY packets for 0.5 second. This would kill all isochronous uplink traffic, but allow all of the couch potato traffic including streaming video to continue to work. You still have to convince Congress that a conspiracy by access providers to deploy such a scheme is worth not enforcing antitrust laws against. Assuming one could get the government to agree that preserving the right of the carrier to block isochronous uplink traffic will prevent, say, Child Pornography or Terrorism, you’ve got a deal with any Congressman willing to take your PAC money. There’s room for creativity here - there’s 136 no lack of willing technologists out there who will creatively argue that there is “risk to the network” if certain applications are allowed to flourish - remember Carterfone and Hushaphone (and remember that in Hushaphone, the courts were persuaded that a piece of plastic on the handset threatened the entire network). Coluccio (referring to Francois Menard’s last post above): Francois, I’m glad you received my post in the constructive manner in which it was intended. “…I would have said that your message had been written by someone who doesn’t understand the challenges of competitive entry. Evolution is the only way in telecom because regulators will not let revolution happen.” My point was this. Embracing the idea that efforts to make the markets for T1 hierarchical and DOCSIS models fair, is, in fact, a means of forestalling their inevitable demise by reinforcing their legitimacy in a way that impedes the very evolution you mentioned. Menard: I fully agree, but if the fail fast model doesn’t work, then the tie’em up model is the only alternative. I have perfectly good models for infrastructure deployments, with real customers. However, these models fail upon presence of anticompetitive behavior that can take many forms. The inevitable demise of the ILECs and of the MSO’s can only arise as a result of the entry of competitors or the refusal of their investors to further bankroll their debt into the future. With no competitors, why would someone find a risk into that? Coluccio: Francois wrote “There are too many jobs at stake …” This gets closer to the heart of one of the more emotional elements of this subject, which I agree is important to consider. But the disruption to jobholders isn’t immediate and wholesale, neither is it permanent for the greater majority of those affected, as some will receive early re- The COOK Report on Internet May June 2005 tirement or parachute on to better things. And it doesn’t begin and end with front line employees who hold union cards and their immediate bosses and middle management working stiffs, either. What the Regulator Will and Will Not Do Menard: My point is that the regulator’s will not do disruptive things. I’ve already seen the CRTC say a couple of times that balancing the interests of consumers and competitors needed to be done against the financial stability of the ILECs. The regulators want them around and will go as far as handing out to them new monopolies as we have seen with the FCC. Coluccio: This form of disruption will also affect, in a direr way and to varying levels of degree, the companies that today represent the paragons of the content and telecommunications industries, as well. Stated in a general way, it also affects major links in the overall supply chains of the vertical group, and where a link is not taken out, entirely, then it will have no choice but to adapt to a set of new rules of engagement, which means that the company either retools itself, and quickly, or it perishes. But the larger number of wage-earning employees will shift to new positions in other companies or industries, and, as usually happens in situations like this one, an equal or greater number of positions will eventually be created that need filling, through the opportunities in new ventures that ensue as a result of the void that is created by the exiting of the old. See my earlier note of today to Malcolm, in this regard. Menard: Given this restriction, and post bubble, I have come to conclude after seeing other (much better VC’ed) competitors fail that we must finance our entry from proceeds generated from the value-added resale of unbundled elements under the protection of price floors. You will notice that our condo fiber entry has also been accompanied by several price floor regulatory enforcement exercises, because if your an ILEC and you’re dominant, you can always bid below cost to kill a competitor then after raise prices to defray these costs. Why, because ILECs can, and in fact HAVE been doing this in Quebec as a result of a mere 150 M$ subsidy from the Quebec government after being unsuccessful in lobbying out the program. Coluccio: Like Gordon, and very likely some other listeners here, as well, I’m puzzled by the $150 Million subsidy to which you refer. Please explain. Menard: The Quebec Government put out a subsidy fund called Villages Branches (Connected Villages) which school boards and municipalities can jointly apply into (they cannot apply without first teaming up) with an objective of financing 66% of dark fiber infrastructures and nothing else (they pay for the remaining 33%) The ILECs tried to make sure this program never saw the light of day and have lost. Consequently, they have been bidding below costs on the tenders issued to this date to make sure that the opportunity for a new build doesn’t fall into the hands of one of their competitors. I’ve been fighting this out at the CRTC for the last two years (with some good success I might add with another victory scheduled for March 9, I’ll tell the list more on March 9 or go hide myself then as a result of my first failure, though I doubt the latter). Coluccio: In recent times some municipalities have yielded to spending money on ILEC outside plant projects, when the ILECs were forced for environmental, aesthetic, or zoning reasons to place their previously overhead-strung cables in underground conduits, or to bury them directly in soil. (This sure sounds like a more environmentally friendly alternative to aerial cabling, doesn’t it?) But not until a lot of horse-trading transpired, first. When the ILECs stated that they would do the burying and or conduiting on their own dime, if they were permitted to adjust their tarriffed cost elements accordingly, hence their pricing and ROI metrics, as well, the municipalities, with I’m sure a little pressure from their state capitals, acquiesced to foot those bills themselves. I guess you could say, then, 137 “Score one for the ILECs.” Is this, or something related to this type of compensation what you are referring to? Menard: Our model of condo fiber on the backbone and WiMax in the access, faces similarly the wrath of forborneconditional-to-the-presence-of-costbased-UNEs ILEC DSL where forcing out the enforcement of price floors is much more complex and intertwined with the provision of local telephony (be it TDM or packet-based). Coluccio: I almost know what you’re saying here, but just in case, would you kindly break that last sentence down for me substituting a lengthier statement for the hyphenated string, and leaving out variations stemming from the word, “forbearance”? Also, please elucidate on your (frequent) use of the term forbearance, describing what you mean by it in the context that you use it. Thanks. Menard: Unlike the FCC which has yet to officially forbear from regulating broadband, but rather has only been nonreceptive to Computer II enforcement complaints, the CRTC has actually forbore from regulating retail Internet and WAN services (in 1999 and in 2000), tough with a pesky little condition which has yet to be fully enforced like Computer II, that is, unbundled network elements being made available to competitors. To date, these UNEs have either been unavailable or the price for them has meant competitors being price squeezed to the point of preventing them from justifying making further investments into their own facilities. With the VoIP public notice, the CRTC will decide whether VoIP-based local services will be subjected to the same price floors than POTS. Therefore, assuming that high speed Internet (HSI) was given away, the bundle of VoIP+HSI would at least reflect the price floors for POTS. Our model of Villages-Branches financed Condo fiber + WiMax in the last mile works, but only if there is a price floor COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA for POTS. The decision for price flooring VoIP to the POTS levels will be made by June 2005 in Canada. On March 9 Francois Menard: As promised: An ILEC making 8 billion dollars a year of revenue caught pants down bidding below cost (i.e. below tarriffed rates) to evict our competitive entry: http://www.crtc.gc.ca/archive/ENG/Decisions/2005/dt2005-12.htm Wenger: Congratulations Francois. A fine victory indeed. What are the implications for the growing community network community? Menard: Thanks, that’s a small reward for countless hours over the last two years at understanding the regulatory framework in Canada. I guess that I’m now entitled to charge as much as a lawyer. If nothing else, this decision will certainly discourage ILECs from lobbying community networks to go in the bed with them to spread the ILEC dominance disease which prevents you from even having the will to consider competing with the ILECs. Deals from ILECs which are too good to be true USUALLY are and often disguise a) the advancement of rate increases for uncapped local telephone services (such as Centrex) and b) result in the eviction of competitors which do no longer have the economy of scale to pursue facilities-based deployments, which b) then result in b.i) ILEC raising prices by way of only offering lit services which are much more expensive than dark fiber services. All of which are antitrust-type plots leading to market eviction of new entrants that fall in the category of criminal offenses. Wenger: Does this simply close the door on some loopholes for anti-competitive behavior, or does this open some proactive doors for them as well? Any word on the Dark Fiber Tariff or how that is coming? The longer it goes, the less optimistic I am. Menard: It means that the public interest will be served by truly competitive behavior between carriers rather than have a special breed of carriers that are not in the business of competing the ILEC. Did you know that the CRTC ordered TCI in Quebec and Telebec and Bell Canada and Northwestel to file dark fiber tariffs, but that in each case, they made sure that the tariff was much more expensive than the cost of construction, even for a single pair of fiber. While true that the CRTC has yet to order TCI to file an IX dark fiber tariff in BC + AB, even if it did, I would bet that TCI would make it so that the tariff would not be financially in reach of community networks. Wenger: On another CRTC related note, I note Canada is beginning a major review of the Commission and the press makes it seem rather ominous. Like kiss the commission good-bye... While I have not always agreed with the speed or reasoning of their decisions, I note the past 18 months has seen a rather positive turn for the better in the manner in which decisions are protecting competition (maybe they’ll actually begin to encourage it soon!). To kill them now seems almost ironic, but the cynical view would say that now they’ve stopped protecting the rights of the ILECS, their value is limited... I hope that is not what you are hearing on the frozen streets up there... (Come to think of it, mine are still frozen too). Menard: Any telecom policy review will be coming out through a finely reduced and well-defined scope for a public notice that will attract several hundred pages of filings which will force the reduction of the arguments to something which is hopefully in the public interest. I would be surprised that Parliament will be as bold as rewrite the telecom act without making this a public process that is accountable. In other words, I now know how to play the game and play it well, and I trust that the ILEC have little to no chances to get a new monopoly in return. 138 I do have to admit that even yesterday, I doubted about winning and I was not overly optimistic ... and look at what came out... the CRTC even said that equipments were regulated and we did not argue that at all at great length. When you’re swimming in blood, they listen to you ... we just need to well articulate why we’re swimming in blood which is easier when you are being foreclosed from competitive entry. This is easier than to demonstrate that you are subject to an unreasonable price squeeze even if such is even more true. Forster: Yes, congratulations Francois. I think you and I disagree on some aspects of what should be regulated, but I support and applaud your efforts to keep the incumbents from using unfair business practices. Coluccio: Congratulations, Francois. Several days ago, when we were discussing your petitions at the CRTC concerning open access, voice and pricing floors, I was unaware that they included Dark Fiber pursuits, as well. I may have missed that point, but Kudos to you, anyway. Your reply to Matt highlighted something to me as though it were an epiphany, which it really shouldn’t have at this late state of the game, but it did. And that is, it should be the regulators’ to see to it that the monopoly holder for a given territory provides its subscribers with the types of services they demand, in this case dark fiber, and not simply the family of services that fit their historical computational methods for determining ROI based on legacy services, alone. Yeah.. yeah, I know. But it did me good to hear myself say it. Menard (earlier) : So just to make sure we do not lose our shirts deploying WiMax, I must fight it out on the DSL and DOCSIS fronts. If I cannot finance WiMax because of the risks, then FTTH is out of my league, at lest for the next two years. The COOK Report on Internet May June 2005 Coluccio:What are the basic capacity utilization and planning assumptions that go into your WiMAX model? Can you share those with the list? Also, are you looking at WiMAX as a form of remedy to satisfy extreme rural situations, or to support mobility and ad-hoc wireless networks and backhaul, irrespective of tele-densities in the regions in which your plans call for deployment? St. Arnaud: I think Cringely’s article is an example of Odlyzko’s price discrimination model in another guise. I don’t one should regulate against price discrimination. There are other effective techniques: (a) Ensure there are enough competitors, so that you get at least one network equivalent to Wal-Mart ( but this is hard to do, if not impossible, in the last mile); or (b) Give the customer control of the network and QoS knobs through ownership or other techniques until they can reach an interconnection point where they can meet with many competitive service providers. (This is happening in the enterprise world - but will that model extend to the home?) Vonage Suffers Widespread Outage on March 4 Coluccio: Here’s a brief report from convergedigest.com on the Vonage outage. Vonage suffered a widespread network outage on March 4 at 2:45 EST affecting nearly half its 500,000 customers across the U.S. Both inbound and outbound calling via its VoIP service, as well as its website and voicemail system, were down periodically throughout the afternoon. Vonage attributed the outage to a glitch in a software upgrade to its network. http://www.vonage.com Coluccio: Glitches happen. During the late Eighties when Signalling System No. 7 (SS7) was first coming on line, entire sections of the US were without service due to a software glitch. Sean Donelan raised an interesting question on NANOG related to how VoIP carriers like Vonage should be regarded in situations like this one. Sean asks: Out, exactly what promise are you given of being able to complete a call? Skypedoes not claim to replace telephone service at this point in time. “After Vonage was successful in getting the FCC to fine another provider, will Vonage accept the FCC’s jurisdiction and comply with all the outage reporting and regulation requirements expected of other telecommunication providers? For example, telecommunication providers have to report outages of more than 30 minutes affecting more than 30,000 subscribers. Vonage’s outage on Friday affected 250,000 subscribers for over 45 minutes. Although Vonage claims its the first outage related to software, Vonage has had other outages. Or do folks expect Vonage to flip-flop and claim they are an “information service” and don’t have to follow FCC regulations.” Vonage, on the other hand, differs in many ways. It claims to be a phone service, and even apologizes for not being able to offer 911 properly. It sells its interconnect to the phone network as a primary function, and so forth. So it is very much like a cell phone carrier. It has no monopoly over customers in any market, so it has no “special” role to serve the public, other than its contractual role. I think that accountability in this case is a good thing. It at least allows the user community to gauge the historical performance of a provider over time, thus enabling users to do a more intelligent selection. When the big fishes become subsumed by the littler fishes, at some point one has to also ask, should those little fish like Skype also be held accountable for their performance? It’s not as though Skype had absolutely NO infrastructure, for the servers and the very code that they use constitute some level of liability that can be measured in kind. What do you think? Oh, I know, there are those who just want to twist the law in any way they can to “get” Vonage and Skype . And they may succeed. After all, the courts agreed that a little piece of plastic on the handset was a risk to the entire Bell System in the Hush-A-Phone case.... an we all know that Vonage and Skype are in league with the terrorists and child pornographers (not to mention the indecent NFL and Janet Jackson). Reed: This is very interesting. It draws an interesting difference between Skype and Vonage. Skype charges you nothing for calls between subscribers. Skype’s ability to connect calls between subscribers is quite robust - even if all of Skype ‘s servers went down for 24-36 hours, you could complete calls (at least if you believe them that information about connections is cached in the endpoints and shared among them for 72 hours). Skype charges you nothing for the software as well. So the outage of interest is Skype “Out” failure. Presumably those interconnect points are few in number, and thus vulnerable. But if you buy Skype 139 So I think Vonage should be treated like a cell operator, more or less. When a cell tower dies (and this happens often enough) are they called on the carpet by regulators? I haven’t heard of this, but I could be wrong. Coluccio: David, since the Internet has subsumed so much of what was once deemed to be critical communications in the past, the greater question should be: Should _anything_ be reported on a mandatory basis, and if so, then why not ISP and backbone provider outages, as well? It’s an old question, I know, but I believe it has more weight today than ever before, due to the shifts that have taken place. And granted, if _all_ types of entities were to be included, then there should still be room for means testing the smallest mom and pops based on their size for exemption purposes, too. Thoughts on this entire matter would be appreciated. Also concerning the cellular ops: It was true until recently that cellular/pcs/paging operators were exempt from reporting requirements, but due to the domi- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA nant roles that they have grown into after displacing a major portion of all landline service activity, the FCC changed its guidelines to include them, as described in the following FCC-issued February 2004 NPRM: mization of route costs through whatever criteria a user selects, also permits the kind of fallback you’ve referred to in an automated way, and this is something that is typically used only by large user organizations. http://hraunfoss.fcc.gov/edocs_public/attachmatch/FCC-04-30A1.pdf As a consumer I do the same thing manually anytime my cable modem service, RoadRunner, goes out on me. Which is to say, I immediately press a few keys and I’m immediately onto my NetZero dialup account, and continue along, working on an impaired basis. from p. 28: ---snip: “Our proposals would require the reporting of outages of at least 30 minutes duration that meet specified criteria. One of the criteria is that the outage potentially affects at least 900,000 userminutes for providers of telephony and/ or paging services (including wireline, cellular-type wireless, cable telephony, and satellite telephony services). Those communications providers that would qualify as “small businesses” are, we believe, highly unlikely to experience outages of sufficient magnitude to meet the user-minute criterion. If they were to experience such an outage, then a likely inference would be that a small number of users had lost service for several days duration, a situation of which we should be apprised. We do not believe that it would be wise to exempt small businesses from the proposed requirements to report outages of at least 30 minutes duration that also satisfy the other proposed reporting criteria (i.e., those criteria that are not expressed in terms of user-minutes), such as the criteria of potentially affecting special facilities, offices, or services including 911) or presenting major infrastructure failures or SS7 problems.” ---end snip Reed: I personally think that the market approach to solving the outage problem solves a big part of the problem, if as a customer you can discover your own outages and switch to a competitor (or use multiple suppliers and push business to those who earn your trust...). Coluccio: David, I didn’t expect quite that kind of response from you, but I’ll take it for its stimulative effect ;) Multi-homing tools, aside from allowing weighted load sharing and per-flow opti- You do of course see the contradiction in your last message, though - inspired perhaps by the irrationalities that exist in the Internet “market place”** - by suggesting first that there shouldn’t be more than one provider, and then stating that a user should be able to detect a failure on its own and switch to an alternate provider. Not to be repetitive, but I think this is fitting here: I recently started a conversation over Skype, then jumped to Cellular when my PC audio bummed out, only to have to revert to my PSTN wireless handset when my cell phone battery died. Allow me to muse here for a moment: *Skype* re: Market Place: Is the Internet as we’ve come to know it a common form of “market place”, yet? If so, when did it finally become one? From its inception, egalitarian modality ruled its existence, only to find at one point it had become commercialized, or so it was told. But its binding customs and cultural mores as they apply to many of the surviving non-Tier 1s and some of their heirs, have persisted and influenced its day-to-day operation right up to the present. So,in a sense, those ISPs that are still the believers of yore might be thought of as the singular cloud your message implies? While the larger BBs and NSPs have led the influence curve toward the market place form of commercialization shared by traditional carriers? Does any of this make any sense in the context of your dual message above? I suppose that even if it did make sense, one would still have difficulty circum140 venting a backbone failure, especially if the affected party were secluded in a rural area. Just some musings on what you said. Forster: Right...And the Internet, web pages, and blogging enable a very easy and efficient reporting system. Coluccio: Right. But beware of the signaling channel that is embedded within your bearer channel. In other words, keep at least one POTS phone in place when it comes time to take the axe to your PSTN account ;) On that note, five years ago I’d have pooh-poohed any notions about a threat to the Public Internet as a whole. Today I’m far less sure. If pressed, I’d have to say that the probability of a system wide failure due to security breaches inflicted by black hats and hostile agents are higher now than ever before. Not a comforting thought, but consistent with the flow of this discussion. Reed: The bigger problem today is having multiple suppliers at all, given the propensity to let big suppliers block new entry via regulation, “national security”, and other games (at the local and the Federal level). Forster: A cable industry VP once said at a national conference, on the subject of getting business from the state governments for fiber, SONET, and Internet services something like this: “We can win in the market but we’re getting killed in the state legislatures & agencies. The ILECs are writing the RFPs for them before we see them and then we have 30 or 60 days to respond”. That comes from long-term relationships, on both institutional and personnel levels, between the ILECs and the government. Reed: I.e. “crony capitalism” or “crony socialism” depending on which party spins it. It’s corruption no matter how you spin it. No different than Indonesia. Coluccio: Interesting observation, David, re: crony capitalism. Prior to the first divestiture of AT&T, The COOK Report on Internet May June 2005 A refutation of Metcalfeʼs Law and a better estimate for the value of networks and network interconnections Odlyzko: This comment is different from the various discussion threads going on, but it bears on some, and on some considered earlier, to I thought you might be interested in this note. The paper is titled "A refutation of Metcalfe’s Law and a better estimate for the value of networks and network interconnections" some notion of network value. A monopolist or few large competitors can exercise considerable pricing power. They have a downward sloping demand curve where the smaller providers are effectively price takers - participating in a nearly perfectly competitive market with near-horizontal demand curves. In a free interconnection model, any service provider I peer with effectively increases my competition so that I lose pricing power. This is because the networks I peer can connect to my customers just as easily as I can connect to them. I may increase the “value” of my network in the sense that I can now connect my customers to more people so that the marginal cost of connecting more people is less but at the same time my demand curve is flattening because I face increased competition. Authors; Andrew Odlyzko and Benjamin Tilly Abstract Metcalfe’s Law states that the value of a communications network is proportional to the square of the size of the network. It is widely accepted and frequently cited. However, there are several arguments that this rule is a significant overestimate. (Therefore Reed’s Law is even more of an overestimate, since it says that the value of a network grows exponentially, in the mathematical sense, in network size.) This note presents several quantitative arguments that suggest the value of a general communication network of size n grows like n*log(n). This growth rate is faster than the linear growth, of order n, that, according to Sarnoff’s Law, governs the value of a broadcast network. On the other hand, it is much slower than the quadratic growth of Metcalfe’s Law, and helps explain the failure of the dot-com and telecom booms, as well as why network interconnection (such as peering on the Internet) remains a controversial issue. From this perspective, I don’t think it is at all surprising that large networks are not willing to freely peer with small networks. This is all sort of last year’s news, however because when the bubble popped we had a whole lot of competition in the network core but a very mixed and largely uncompetitive market on the edges. We’re now in a period of major consolidation with a few large victors buying up the core so it appears that we’re headed for a whole new game. However none of this answers the question of how you maximize the value of the network to society? Odlyzko: Sorry if the paper was not sufficiently explicit, but the viewpoint in the paper was definitely that of the customer. There is extensive discussion of the value to an individual customer, and total value being the aggregate of the individual valuations. The issue you raise is that of how the service providers can extract enough of the value the network provides to the customers to pay for constructions and maintenance. That FULL PAPER AT: http://www.dtc.umn.edu/~odlyzko/doc/met- is a key question that is outside the scope of the paper. calfe.pdf Forster: Interesting paper. Retzer: You make some very good points, Andrew but also miss a key issue in my opinion. First off, I must say that I Einstein’s Relativity clearly superceded Newtonian physics, wholeheartedly agree that not all connections are equally valu- but within the domain of non-relativistic speed and mass able especially at any given point in time. I also agree that Newtonian physics is quite accurate. Maybe we should look at locality is an important factor. However, your paper misses the Metcalfe’s and Reed’s ‘Laws’ as being accurate within certain mark in my opinion by not identifying from whose viewpoint is domains based on the homogeneity of the users but inaccurate the value measured? Based upon the general thrust of the paper as the population and heterogeneity grow. and your other work, I have to assume that you are looking at the question through the eyes of the service provider who In the early days of the Ethernet and the Internet the networkwishes to monetize the value of their network. This actually ing community was quite homogenous. In the Universities confuses the issue because the value to the service provider is they were collegial; in the workplace people were cordial and not the same as the value to customer or the value to society. professional for the most place (flame wars a possible excepA network infrastructure provides customers a great deal of tion). As the network grew within these fairly homogenous surplus value beyond what they pay the service provider, and communities Metcalfe’s & Reed’s Law could well have held. provides society as a whole even more surplus value than it Social mechanisms (peer pressure, etc.) helped keep behavior does the direct customers of the network. consistent with group norms. As the number of users on the Internet approaches a significant fraction of the population of The value observed by the network provider, however is a the earth the homogeneity declines and peer pressure is ineffunction of supply and demand that probably has far greater fective. impact on a service provider’s willingness to interconnect than 141 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA maybe going back as far as the Manhattan Project (Sandia Labs), and then the ‘cold war,’ the Bell System held a very special position in the eyes of many government officials that was second only to that of religion, which effectively made Bell exempt from most of the usual separations between church and state. speak of has survived to this day, and why a special regard has always been held for Mom. Some roots continue to grow on their own, even after the tree is chopped down and used for kindling. And so it goes. all the London terminals. Skype in Hotspots As the sole provider, hence, guardian, of the national resource known as the telephone system, Bell found itself involved in the inner workings of congressional committees performing risk assessments and making recommendations at first in connection with the Soviet nuclear threat, and then the nation’s top security agencies and advisory positions like no other private company of its time. Hassinger on March 3: Found this on Ben Hammersly’s blog: http://www.benhammersley.com/weblog/2005/03/03/ snow_and_.html Coluccio: Yes, the same thing jumped off the page at me when I read the release, too. I wonder, even, if a hot spot even needs to be a participant and announce its availability for Skype access. It’s a neat, timely promotional effort, especially at a time when most users, as the second piece noted, don’t have -enabled handsets, hence providing for a built-in buffer to over-usage. How sustainable this could actually be when users become equipped, given the scaling constraints that access points demonstrate with regard to growing numbers of voip - as discussed here very recently - remains an issue I’d be interested in hearing more about. These circumstances engendered the lofty, near-sacred image that Bell has grown to enjoy, which has survived the integrated circuit chip, Arpanet, two divestitures, the bubble, and now, what we like to think of ourselves living in since being liberalized by the power of IP - a new age of enlightened reason, as well. It’s no wonder that the cronyism you “Meanwhile, one terribly interesting thing to arrive in the Press Release Inbox this morning: have signed a deal with Broadreach to allow to be used free over the 300 or so ReadyToSurf wifi hotspots in the UK. Broadreach’s ReadytoSurf(TM) network of locations has a combined footfall of over 1bn per year and includes major brands like Virgin Megastores, Eurostar, Travelodge, Moto, Little Chef, Virgin Trains, EAT , Choice Hotels and Quality Inn and major railway stations including 142 You’ll still have to pay to get web and email, but the firewall is unlocked for Skype traffic. You hear that? That’s the sound of the future.” [Editor: Frank did hear more. I have however moved the result to pages 106107 above where it follows another conversation on security issues.] Summary Conclusion March 11-17, 2005 Instant Voice, Chat and Messaging Motivate Early Adaptors to Rethink What it Means to Be Connected Highlights Editor’s Note: One of the enduring lessons that I am carrying away from a three month immersion in SKYPE, SIP and VoIP is that while it may yet be a long time before voice leaves the PSTN, the richness of opportunities to communicate is becoming so stunning that the very multitude of new options will lead to difficulties. It starts from the very scarcity model of the phone company where we communicate one-on-one at the sufferance of the commanders of the “intelligent network,” where our identities as phone numbers are on view to the world unless we pay to keep them hidden. What is emerging is a much different model where if we don’t take more direct control over our identity and reach-ability we will be overwhelmed. The new tools can be used in many new ways as the discussion with which we shall close this symposium shows. What is uncertain is how far and how fast this perceived need for identity management will spread. The user communities involved with tools like Skype will need to work out ways of using the tools that are acceptable to others. Skype had better listen carefully to its users and incorporate within its product the ability for users to control the extent to which they may be reached and the conditions under which they may be reached with these tools. And above all Skype had better look at engineering barricades against spam. This is an early adopter, leading edge experience to which the Skype community, the handset makers and the phone companies at large need to pay careful attention. COOK Report: On March 11, James Enck posted a short essay he called Skam on his blog http://eurotelcoblog. blogspot.com/ Enck: I’m going to coin a phrase here, I think. Anyway, Skam = Skype spam. A friend received the following message yesterday: "Dear friend, I represent a consulting company from Latvia. The company is one of oldest in renewed Latvia. I have a proposal concerning legally reducing income taxes. You would be surprised about the results. Agents interests are appreciated." Skype. I am also a bit concerned that, as it gets bigger and better, it is also more attractive to those who have less noble intentions. On March 15 COOK Report: I have been a little concerned about whether my repost of James Enck’s Skam here a few days ago rubbed salt in any wounds. It was not my intent. This morning I opened up Skype to find that Stuart Henshall had initiated a group Skype chat for 46 people last night while I was offline, in order to discuss SkypeIn. The entire archive appeared in the client, and contained some interesting reactions, as well as some baffled ones. It seemed that some people were confused and perhaps vaguely resentful of being dragged into a group chat. Several exited immediately. I instant messaged Stuart and found this Blog entry below in reply. I certainly don’t blame Stuart, because I like well-intentioned enthusiasm. But it made me start to think. Is it possible to repeat this process with people outside your contact list? I assume so, but I’m not going to try it. How about a Skambot, which indexes Skype users from the directory, enabling Skammers to set up unsolicited super-sized group chats like the benevolent one I experienced - but with the message being Latvian tax strategies, or worse. What defense do users dragged into this sort of thing have? How about unsolicited file transfers of advertising, porn clips, or worse? Remember, I wasn’t online at the time the chat started, but still got all the messages when I booted up today. Yes, I have a right to exit, but do I have a right not to be brought into it in the first place? My feeling is that I don’t, not at present anyway. Since I was not directly involved, my reaction was that James’ original post was a reaction to an unexpected invasion of presence and not an accusation of spamming. Still, this seems to be a case where putting new tools in the hands of a wide group of users will cause inevitable bumps that require thinking through as Stuart has below. I love Skype as an application and a force for keeping the telcos awake at night. I admire the spirit of innovation that Skype is pursuing. I am amazed at all the activity being inspired around 143 I think it is a good reply. Stuart also took me through some privacy settings and chat settings that are not available yet on the MAC Skype client - I therefore have a bit of a better understanding of what went on. If I understood Stuart correctly a Skype megachat (limited to 50 people) is like setting up an IRC channel. In http://www.skypejournal.com/blog/ archives/2005/03/buddychat_insig.php Stuart wrote Henshall: James Enck is concerned about Skype Spam. I was wary then intrigued by his response. It just proves that you experiment and learn. When I first saw his blog I thought: “How could you”? Then I asked myself: “What needs fixing?” I have a suggestion and a solution that could further improve Skype. It involves more control over Skype Mega Chats. When SkypeIn launched last week I saw a timely opportunity to share this news with a group of buddies all of whom use COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA and understand my interest in Skype. According to James, I shared it with 46. It’s called a group chat session (Skype MegaChat!) and you select who you want to add and hit start. Did I become in James’ words a “skammer”? I don’t think so. Still the result led to new thinking. Rethinking Skype Mega Chats In retrospect it was a little like starting a group e-mail list without enabling others to opt-out until they got the invite. The option always existed to “leave” and some did. However, I think for the majority it was their first experience with a Skype Mega Chat, and it caught them off guard. It surprised me a little too. Some thought it neat, others left and some were frustrated with the exit process. To leave a group chat you must hit the “leave” button, closing the window isn’t enough. However, this isn’t where the real story is. The opportunity exists for Skype to insert another level of chat control. This option would automatically refuse chat invites where more than x people are involved. Or said similarly, “limit” my chat participation to chats containing 4 or less. The mega group chats can then be always “blocked” or put into a call/chat list context that doesn’t pop-up/ring buzzers, etc. This would be useful. If you have ever had a group chat going while being on a SkypeCall and haven’t turned off the message “whistle” notification you will know what I mean. Creating two chat types would add new utility. There are pervasive chats that many want to leave open, similarly the head of a small company may start a group chat to provide an update. Traditionally this may have gone by e-mail, some are even broadcast phone messages. Sending it via Skype means you will get it whereever you are logged on. That’s interesting in a mobile, multi-device world. It seems like it may follow you better than e-mail and, along with presence, be more timely. The Skype mega chat is similar to an e-mail list. However, to give it the priority of a oneto-one chat is wrong. One-to-one chats are invasive, less so than a call, and more so than just leaving a voice message. My learning is that group chats are interesting and very powerful. I didn’t understand the power that “initiating” one has. Rather than have reservations, I’d like to be able to share more news with my buddies. However, it doesn’t deserve the visibility of a one-to-one chat, or the chat that invites you into a conference call. For that we need the fix suggested above. Thus, let’s not conclude that we have a spam problem. However, let’s manage to enhance the experience. I could indeed start a mega chat with unknowns however, only with people that accept chat messages from people not on their buddy list. Potentially I could ring and call and do all sorts of things to capture the list. However, it all falls down when Skypers set their privacy options to reject calls and chat from users not on their buddy lists. At that point you must ask for an authorization to interact. Coluccio: Interesting, Gordon, Stuart and James. Allow me some musing for a moment. One of the notions I had, several times when reading through your last message, was the level of intersection that exists and the possible redundancy that is taking form, as well, between Skype and regular forms and uses of e-mail and IRC. Especially when the feature capabilities of one of them has to be analogized in order to describe what is happening in, and the nature of, the other. System” [Editor’s Note: This system uses the federated P2P architecture developed by Nimcat Networks in Canada and similar to Goroshevsky’s Peerio both of which are discussed in the interview with Jame Enck on pages 57-65 of this issue.] Powered By Nimcat, The VentureIP System Enables SMBs To Easily Install and Operate A Full-Featured Phone System By Simply Connecting Telephones To Their Network TORONTO (March 14, 2005) - Aastra Technologies Limited (TSX: AAH), a leader in enterprise communication access products, extended its industry leadership today with the launch of VentureIP - the first enterprise-class, P2P (peer-topeer) IP-based phone system. Powered by Nimcat Networks’ nimX software, the system enables small-to-medium business (SMBs) to install, operate and manage a full-featured phone system by simply plugging the VentureIP 480i telephone into their local area network (LAN). The system scales on a phone by phone basis and can be connected to the local PSTN via the VentureIP Gateway. Meanwhile here is another P2P solution, this time for enterprises. Or so says the PR: “We are pleased to offer the first IP-based, P2P system in our industry that meets the enterprise-class telephony needs of the SMB market,” said Hung Lam, director IP product portfolio at Aastra Telecom. “When a new VentureIP 480i telephone is added to the network, it independently recognizes all other VentureIP terminals and gateways on the network and instantly forms a trusted, virtual exchange - with the ability to interact with its peers and to connect to the PSTN. When a user connects a VentureIP 480i telephone into their data network, the device automatically configures itself, allowing calls to be made and received without any additional, complex setup or centralized server equipment. Additionally, through its connection to the LAN and the Internet, users have Web-based access to the VentureIP system to better manage their set’s options, as well as access to future software enhancements.” “Aastra Launches VentureIP - The First Enterprise-Class, P2P, IP-based Phone [Editor: BIG SNIP of the remainder of Astra Tech announcement.] At what point are they truly redundant? When sufficient scale in membership to Skype is reached? Or simply when enough of one’s own, personal, trusted parties joins onto a service, where no other parties still matter? I never said this would make any sense. Just musing, as noted. But the notion did crop up, just the same. 144 The COOK Report on Internet May June 2005 Henshall: Please tell me why I should get excited about Aastra? http://www.aastra. com/enterpriseip/pro_228.asp What I see is a telephone with an oversized LED. I don’t see an instrument of the future here. While it may eliminate some back end elements and provide a lower cost, there will be no “joy” from consumers learning to use this phone. There is no story about changed work practices. It’s productivity that is the real opportunity here, not cost. The future is in “Presence” and tools that enhance conversation and communication. The handset I see here isn’t even cordless. I really must be missing something. Separately, I added to my desktop a “telephone” today, actually it’s a VoIPVoice product I said I’d test and learn from. www.voipvoice.com. See the CyberPhoneK. It integrates with the Skype API, it’s fairly basic $50.00 retail. With SkypeOut and SkypeIn it does more than any PSTN phone. However, having it on my desktop is a retrograde step. I’ve been using a Bluetooth headset. It’s paired with my laptop when I’m home and with my mobile when I’m out. My hands are free to work and text, no need to hold a phone. I can even go and make coffee. There are usage occasions where a CyberPhoneK has real merits. I’ll save that for another time. Aastra’s underpinnings may be neat and revolutionary; however “users” adopt new tools and demand them when they bring delight, provide advantage, etc. The Aastra Venture IP phone looks many times more expensive than “software” and a longlife Bluetooth headset and a good mobile PDA. I know which way I’d be working to take a small enterprise. I’d love to be convinced and shown why this is “NOT” yesterday’s product. Coluccio: Stuart, thanks for responding to both messages. My reporting on Aastra was simply that: reporting. I was simply publicizing it for the list. If I’m reading you correctly, then I base my world of telephony contacts on who it is that I find in “my own” network, then by definition I either find myself tacitly or deliberately excluding anyone who is not in my own network. And if I don’t outright exclude them and choose to contact them through ordinary means, then I am at least more likely not to come across them as a matter of routine, or even casual, encounter as I would if they were a part of my own network. If those same excluded parties, in turn, are a part of their own network(s), my chances of contact with them is doubly or “nth”ly reduced, depending on how many “their own” networks exist. Except that, if all of “their own” and “my own” networks were made to be compatible with one another, then this balkanization would not so readily occur. Without such compatibility, Reed’s Law, not to mention the benefits to be derived from sharing ideas on a global scale, goes down the tubes. I was discussing this eventuality this evening with a colleague in Boston during a very long Skype-enabled conversation, and I suggested to him that the problem I’ve posed here could become exacerbated if there were five or six more popular variants unleashed. His comeback to me was, “Why five or six, and not fifty or sixty?” To which I responded: “I was only talking about the next couple of months.” Davis: I remain to be convinced of your premise (RE your remarks on Skype vs Aastra), that being that you know the complete story on “productivity” and “changed work practices,” “real opportunity,” what will “advantage” an unlimited set of users. If the butter on your bread is spread by the knife of “presence,” “always-on availability,” your Bluetooth headset to softphone, then more power to you. Such is a different statement than saying what is best or better for “everyone.” In my world, such is level of the first line of control and scope increasingly filtered “presence,” mitigated 145 demanded at the supervisor. Span of work require communications by priorities, up the chain and across scope focii. Human transactions are touted in “popmanagement” and “pop-psych” literature as equivalent to worker “empowerment,” but only for the dispossessed. Human transactions are a variable overhead cost that must be run through the calculus of cost/benefit analysis once human organizations reach a size of more than 50. Just as industrial design and engineering has deployed statistical analysis on product quality to find that the overwhelming preponderance of error was at the human-to-line and human-to-product interaction nexus, so also is it for interhuman communication. Ninety percent of human speech is not about the content expressed but about all sorts of other primate agendas, including, but not limited to, posturing for rank, contentiousness, turf, “blue sky” attention seeking, scape-goating, and just pure “keep alives.” Those are calculated on the cost side of the cost/benefit calculus. We develop electronic work flow and electronic document sharing to eliminate those long and personnel cost expensive “reporting” and “project management” conference calls. The margin pressure in the global economy just won’t afford them. For me, the future is about “Choice of Tools.” As we move from traditional labor and traditional life long job skills and employer homing, it will be up to the worker to select the Tools and how and when to use them to set marketability. More choices and the ability to choose among tools are what “empowerment” means. More choices do not make for better decisions, of course. And the market is the ultimate judge. Savage: Let me de-lurk here for a brief observation. The underlying assumption of 20th century telephony and later ‘net connectivity is that ubiquity is either the norm or the desired goal. [Insert historical observation/rant about Bell System propaganda value of universal service here.] In fact I suspect that most people COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA have little interest in connecting to the vast majority of others on a “ubiquitous” network. Moreover, interests in connecting are asymmetrical. The group of parents on my kids’ soccer teams need to be able to call each other. I want to be able to call tech support for the products I buy, but I do not want to hear from the company’s telemarketers. Etc. My not-terribly-well-formed hypothesis is that if establishing connectivity to selected groups of people were easy and cheap, what would happen is the creation of islands of open connectivity linked to each other and/or the world by non-transparent “gateways.” Think AIM buddy lists. Once you are vouched for by someone in the group, you’re OK, but until then you’re out. This absolutely interferes with ubiquitous connectivity. My suggestion is that this is not a bad thing. To the contrary, it is perfectly rational for people to value connectivity to certain groups more than others, to value inbound versus outbound connectivity differently, etc. Skype et al. are permitting a certain natural experiment of this phenomenon in a way that the ubiquity model, imposed via the command-and-control Bell System, did not. Hertz: Chris Savage’s suggestion about connectivity is absolutely consistent with the idea of separating the ownership and operation of the pipe from the pipe contents, and moving the network intelligence to the edge. The pipes need to be ubiquitous but the connections do not. There is no incentive for the pipe operators to know anything at all about the pipe users (other than the flow parameters needed to size the pipe). Users (commercial, individual, enterprise) are free to set up whatever connectivity they choose. Or not? Choosing How to Connect with Each Other Becomes Critical Jere Retzer: People want to connect to those whom they want to connect when and how they want to connect and they don’t want to be bothered by those who they don’t want bothering them. That’s why I think we may be seeing the death of the public telephone directory as well as public telephone numbers. Moving from the confined PSTN to VoIP really opens the door to changing a bunch of stuff, in my opinion. That’s because of the automation that makes it really easy and affordable to annoy (read SPAM or SPIT, whatever we’re calling spam over VoIP these days), as well as a bunch of desirable ways and reasons to connect people. Oregon, for example had a “do not call” registry for telemarketing years before the feds. I’ve belonged to it and like it a lot, even if I had to pay $6 every couple years for the privilege - just wish there were a way to ban the political calls during election season - now THOSE are annoying. My e-mail, on some accounts however is just a disaster of spam - three fourths of the stuff I get on one account is spam. Fortunately, I’ve gotten pretty good picking out the junk just from the return address and subject line but it is still annoying. I cannot imagine however, my phone starting to ring off the hook at night with robot calls inviting me to buy Cialis. The fact of the matter however, is that all it takes is some free gateways from the IP to the PSTN world to make that happen right now and once that home phone is VoIP then nothing currently prevents it. Once these spam phone calls are as cheap as spam e-mail the spammers will decide that one hit in 10,000 is well worth it and look out. I sometimes have a hard time figuring out exactly what the big deal is about privacy these days - really seems people are getting carried away worrying about stuff. The other day I got an e-mail, not an urban legend reporting the shocking information that if you type your home phone number into Google, it will offer your name and a map to your house. Same day, I heard a bunch of news stories (as well as ads for shredders) about identity theft with cautions to “shred everything with your name on it.” I felt like wrapping up my Qwest phone book (which I now use about 2-3 times a year) in plain brown paper and delivering this 146 “shocking exposition of private information, no doubt with full knowledge and consent of the Oregon PUC” to the local TV station news department to see if they got the point. A lot of stuff we used to never worry about is now suddenly private information to be protected with all your might. Seems weird. Automation is, of course the reason why people ought to be concerned because it once cost significantly more to indiscriminately market to or use someone else’s identity inappropriately than it does now with the Internet because all of a sudden the price to bother or steal from you has dropped to nearly nothing. I don’t think master directories, such as patient directories are the solution either. Do you have any idea how lucrative a target those servers that have your email, phone and credit information are to people who want to abuse it? Thing is, the more information about you collected in a directory the more valuable it is to people who lack the integrity to honor your privacy and your desire to protect your stuff. We want to communicate with those we want to communicate, however on our terms. We all of us have many different persona - our work, family, hobby, community, various groups, etc. We have different expectations for these different situations as well - when at work I expect certain kinds of contacts but when walking down the street the shields are up. Our “phone system” needs to adopt this sort of paradigm as well, in my opinion. What is a phone number if not simply a pointer to you in a particular persona? There’s nothing magical about 10-11 digits. In fact, that is not particularly secure because it is not a lot of entropy. My work number points to me at work, personal cell to my personal persona, etc. So, for those who have a hard time imagining this P2P phone system that seems to be rapidly evolving in Skype. I say maybe this is a good thing that business, and Skype needs to figure out as well. Suppose I had a business identity or some means to identify business calls and some means to identify personal calls that I want to receive. That would be pretty sweet. I’m not sure how The COOK Report on Internet May June 2005 it would work - maybe the traditional PBX becomes an LDAP database with employee names and business contacts. Not bragging, but I have hundreds of business contacts currently unknown to our phone system and it sure would be nice to have them automatically available to my phone - sort of caller ID of the future. Same thing with my personal contacts but that’s a directory most probably I would like to keep private. Needs thought, but this or something like it is where we need to go with enhancements like if I get a return call from some entry not currently in the book it pops up return call, but if it is an unknown and unverified call it gets screened. For those who say: “You’re using your employer’s systems,” and “You should not be using it in unapproved ways or using their network for personal business.” I say: “Yes, you’re right.” However I haven’t laid similar restrictions on my employer effectively using my personal networks at home, etc. I’m not saying we should abuse our employers’ resources. We assuredly should not but the line between personal and business life is getting fuzzier every day. The majority of sales people in this country today probably telework from home at least some of the time. A lot of us tote around extra devices and phone numbers to keep up a distinction between work and personal life that could just as easily be handled in software. Our employers all have a vested interest to improve our personal productivity and this approach might help. Phone numbers, as I say are just pointers. So, to answer Chris (and I apologize for the long rant), I think we really do want to be always on, and always connected but under our own rules. Thanks for your time. Skype In Looks to Be an Important Intersection Between Skype and SIP Editor’s Note: SkypeIn announced on March 11 and SkpyeOut now a million strong rests on an agreement with iBasis to use SIP to interconnect calls with the PSTN. Shockey: (Please see Stuart Henshalls Interesting blog entry) http://www.skypejournal.com/blog/archives/2005/03/ skype_strategy.php#more Henshall: What should Skype’s strategy for SIP be? With the beta launch of SkypeIn many Skypers are getting a “traditional” phone number. That number is also a SIP number. The Skype interconnect (see iBasis below) to the PSTN required working with partners that could “connect” Skype to the PSTN. Thus all SkypeOut and SkypeIn calls use SIP to connect. This is a question piece, looking at whether Skype should make SIP numbers available to SkypeIn buyers. Then connecting to any SIP phone from Skype is free. (See also Skype and SIP?) Consider -Should Skype offer Skypers their SIP number at a small price premium? This would immediately separate them from almost everyone else outside FWD and PhoneGaim. Vonage’s ATA boxes are locked, CallVantage the same, etc. Even most of the “new” softphone players (TelTel, Teleo, Damaka) are “locked” even when they claim SIP. What are the strategic implications for Skype and competitors if Skype opens up SIP? As a user I could use Skype “supported” SIP features. That may mean I can ring a Wi-Fi SIP phone device, I may find it hard to scroll my buddy list or obtain the same voice quality as I get with the Skype experience. Would this allow me to receive “securely” (I don’t buy the security argument BTW) my SkypeCalls on a corporate SIP device or SIP softphone without upsetting my IT department? If so, this would be pretty cool. At the moment unless I have call forwarding, (or Vonage already at home) I can’t enable my home landline to ring the office. Wouldn’t SIP enable you to ring your Skype almost anywhere? While the call quality would be inferior... the sound would always be better on a Skype soft client or via a Skype enabled hardware device (the approval and integration process is real important). 147 What if Skype just offered SIP numbers with SkypeIn? Would that make them quickly the largest SIP community and deployment? Could Skype then “shape” the direction of SIP and thus telephony. What happens if Skype introduces their own version of SIP (Microsoft vs Sun on Java)? Does Cisco fit in the Skype SIP plans? It’s a fact that Mike Volpi from Cisco sits on Skype’s board. We have examples of both Linksys D-Link router solutions. Jeff Pulver noted back in September last year that they could work a proprietary form of SIP. Is it appropriate to guess that the SkypeIn system now adds @skype.net to all Skype names to create the Skype SIP identity. So I’m stuart_henshall@skype. net or something similar for them to provide the interconnect. That registry then opens up a number of new opportunities and the ability to substitute “corporate” e-mail addresses for the Skype created dummy address. Thus this would enable connections to corporate registries, etc. I’m not the telecom / SIP wizard, I just write scenarios from time to time. Could a leap take us to using the organization’s SIP server to register “internal” Skype clients into the Skype network? All connections outside the firewall would be via the SIP server, and thus the enterprise would have the “security” and functionality of Skype inside. Perhaps complements of Cisco? My real interest in this line of thought began with creating an even larger market for Skype-related hardware. Making hardware that is both SIP and Skype compatible will further accelerate Skype growth. It creates a larger pool of hardware that “works” with Skype while sucking more people into the “Skype” experience. The hardware that Skype approves can then focus on better sound, experience, etc. Done correctly the price premium would be for Skype approved hardware. Interestingly “consumers” could be the one’s paying the royalty via their service agreement. I don’t think Skype began with a SIP strategy, in fact, many press releases complained that COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Skype wasn’t SIP-based in a “how could they” approach. Now in a twist of fate, opening and driving Skype/SIP could bring in many new developers and systems suppliers. It could also prove that Skype has both broken SIP and perhaps ready to press it’s own form forward. Notes: I blogged 5 million Softphones when SkypeOut was launched [on May 18, 2004 ]. They are over one million SkypeOut customers now. Alec Saunders succinctly put it that they were distinctly breaking inter-op [by not becoming completely SIP compatible]. At the same time Tom Keating asked whether Skype had the capability to “skip the SIP bandwagon.” Will Mr. Blog get his wish? “iBasis is a first rate partner in delivering interconnections and handling large volumes of Skype’s Internet telephony terminations to the traditional PSTN network,” said Niklas Zennstrom, Skype’s CEO and co-founder. “We appreciate iBasis’s expertise in delivering innovative new solutions and efficiencies, and we look forward to continued joint success in expanding the consumer adoption of Internet telephony worldwide.” These distributed facilities provide support for multiple protocols and devices, network-to-network SIP authentication, and trans-coding capabilities using SIP proxies, Session Border Controllers, and trans-coding systems within the network ingress points. Jere Retzer: One interesting wrinkle on the linked article is that SIP is really just a way to set up the call so that if both ends have Skype I would think you’d be able to use Skype codecs and other features. Schulzrinne: Which you can already do today. There are no “Skype codecs,” just codecs that Skype licensed from GlobalIP Sound. SIP will happily negotiate them between consenting applications. Shockey: Exactly. My licensed XTEN client can support ILBC very well. SKYPE as I have mentioned here before is a fine application but its still a “silo” just like AIM and Yahoo IM, etc. SIP is a global service based on open IETF standards. The SKYPE application user interface is first rate and very easy to use the Skype team are fine applications developers. As a certifiable SIP bigot I still wish them well. They have demonstrated beyond a shadow of a doubt that voice is still nothing more than a bucket of bits. Now add SIP CUA functionality to the Skype user interface and let me buy a service ...aka sip:richard.Shockey@skype. net add H.264 video along with the free P2P and I’ll be happy to open my wallet. The theory of SIP integration should fit in nicely with Skype’s intention of having a “Chinese menu” of incremental for profit options like SkypeOut and voice mail, etc. There is only one problem with that theory … will the new Microsoft ISTANBUL Messenger client crack open the SIP market by putting a real functional SIP CUA on the desktop. Statstny: Funny, [looking at Stuart Henshall’s ideas for Skype and SIP above]. I had the same idea yesterday when I tried to find out a way to point to my Skype name in ENUM. The easiest way to show this would be to set up a SIP proxy skype.net with sip: fordprefect@skype.net running a Skype client via the API. Since Skype is an NGN, this would of course be an SBC. Skype could do this of course natively and as Henning pointed out, Skype is using open codecs, so the media stream could go direct. Remains the encryption problem […] Coluccio: I was waiting for someone to mention encryption, Richard. Thanks. And this would also raise its head upon attempting any gateway function between dissimilar P2P variants, as well. Would it not? Since my initial post on this subhead, many interesting concepts and individ148 ualized views have emerged. And although it was not my intent to challenge the notions that have been aired here, some good stuff has come out of the discussion, just the same. What do you mean by an “open” codec? Isn’t the codec that Skype uses on the pricey side of proprietary code? Or, am I not reading what you’re saying; the way you intended it? Geddes: I notice Richard’s comments that Skype is a silo whereas SIP is open. Richard’s thesis is more openness equals more value. I don’t see things that way. At the risk of invoking the “evil bit” April Fool’s joke, you can broadly class connections between endpoints into “wanted” and “unwanted.” The value of a communications system to the users is the value of wanted connections, less their “toll” cost, less the expense of handling the unwanted transactions. (All abusive calls are unwanted, but not all unwanted calls are abusive, so these issues are not synonymous with spam or criminal activity.) Today, SIP-based solutions and Skype broadly offer the same up-front communications value of a duplex audio stream, presence and client features like history and voicemail. Application-layer innovations can be implemented either way, and whilst SIP might be more adept at accommodating unexpected usage, this doesn’t seem to be a significant factor today. It’s easy for two users to adopt either system, and neither is much used like the PSTN for soliciting strangers; calls are within defined groups. So the “application value” appears similar. Both have a zero per-call toll cost (onnet). The differentiator in user value therefore is in the unwanted connections. A totally open SIP solution with a “naive” ENUMtype call routing solution will result in the mother of all voice spam festivals. This is likely to swamp any positive benefit from increased reach of addressable endpoints. Richard is already working on adding pointer data to ENUM as the first steps to creating identity meta-data you The COOK Report on Internet May June 2005 can use to filter callers; there remains a lot of work to do. Note that the alternative -- Carrier ENUM -- is again a closed “members only” system. Skype’s closure is therefore a feature, not a bug, in the absence of a scaleable public identity system to complement an open protocol. If you look at some of Stuart Henshall’s or James Enck’s recent postings, you will see how important are limiting communications and connectivity. The features they seek are to *not* participate in large group chats, or *not* to receive long voicemails. SIP and ENUM attempt to import the best of SMTP, HTTP and DNS. Unfortunately, they also import the worst aspects of those, and without regard to the fundamental differences between media they carry. Unlike e-mail, you can’t use the message content to filter voice communication. The interruption -- and cost of the unwanted call -- happens up-front. SIP and ENUM cannot survive as currently constituted by copying the open model of e-mail. Today they can only prosper as hidden components of closed systems. In the past, the value of telephony came from connecting people over scare connectivity. Now, the “connecting” effort over and above the packet data costs are negligible. In the future, value will come from *inhibiting* connections over abundant connectivity. The paradox of the paradox of the best network! (See http://www.telepocalypse.net/archives/000407.html). You make more money from stupider networks by throwing stuff away intelligently. How this might impact the end-to-end principle is a post for another day. None of this changes even if you move to a mesh utopia with no service providers. Coluccio: After reading Martin’s and several of the others’ views on what constitutes an open or closed network I’ve concluded that: -- no network is entirely open, since, in the extreme, the only natural cord to the mother ship that we’re born with is cut within the first few moments of our new lives on this planet, which was soon followed by a smack on the ass just so we wouldn’t forget it. And, -- very few networks are entirely closed to the individual who’ll spend the time and effort to become included as one of its members. I found myself musing over what IT managers who’ve suddenly found themselves in charge of voice services, as if voice weren’t data, too, in the world at large would have to say on the subject. I thought of them because those are the folks who pay the bills that I send out once a month, which allows me to spend extraordinary amounts of time talking over Skype and writing sometimes profusely to this list and elsewhere about things that my clients should be doing but aren’t. Of course, enterprises each have their own forms of open and closed networks, too. On the side of open there exists, variously, the Internet, the PSTN, or simply the horn. The closed options are --with “closed” to be taken with a grain of salt-- private lines, intranets, VPNs, and insulated networks that are entirely cordoned off, often owned outright by the enterprise, from even their own internal intranets, with the latter serving to satisfy situations requiring the “utmost” in *secure* communications. And some closed systems take the approach of using lines and switching facilities on the PSTN and the open Internet and encrypt everything so as to be unintelligible to all but a privileged community. Stastny: I think there are some flaws in your argumentation. One is that you mix up openness of a protocol with security and privacy issues. Skype does not provide you with any security from other Skype users, so if anybody is finally in Skype you will get called by anybody again. Hiding of contact details is also not solving the spam and spit problem. I am just reading Rick Whitt’s (MCI) papers on horizontal layering and regulation. http://global.mci. com/about/publicpolicy/presentations/ 149 horizontallayerswhitepaper.pdf I think you are also trying to solve a problem in one layer in another layer, namely a problem of the connectivity layer in the identity layer. As we talked yesterday: SIP and Skype are providing connectivity end-to-end via the transport, but it is finally the problem of the end-users to solve their identity problem between them. This is especially true in P2P systems where nobody is there to solve this problem for you anyway. Richard Shockey wrote: SKYPE as I have mentioned here before is a fine application but its still a “silo” just like AIM and Yahoo IM, etc. SIP is a global service based on open IETF standards. Retzer: Well, partly just to be contrarian, but also because I’d really like to see some real progress in VoIP please allow me to pose an alternative view in response to Richard. I also believe and fully support the idea of open standards. That said, we see in some cases (such as MS Windows) victory by proprietary over open. Why is this? In my reading on the network economy, one commonly proposed thought is that in an industry characterized by network effects there is a tendency toward monopoly or very limited oligopoly in particular niche markets. How do you reach a global consensus on how to re-engineer the phone system when a number of the key players have a substantial vested interest in the legacy PSTN where they still make far more money then they will probably ever make with IP voice. “Why,” I can just hear IBM execs asking in 1985, “would we want to put a lot of money into developing this PC thing when we have these wonderful mainframe computers?” Success today is at war with success tomorrow. Then along comes a small company with a good idea and product that proves unexpectedly popular (albeit proprietary) that takes off. All of a sudden you have success emerging from the endless de- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA bates. The world loves a winner. An empire is born. One notion that keeps coming back to me as I think about our current market is the idea that one man’s barrier is another man’s opportunity. In our world, if the incumbents throw up barriers to competitive, independent VoIP services - arguably easier if the competitors are using open standards - they are in effect creating a huge reward opportunity if someone finds a way around their Maginot line. Statsny: That’s exactly what happening with SIP and Skype, and they also will kill in the same way ENUM, only to loose it to Microsoft. Retzer: We seem to see over and over success in the arena by those who know how to change the game, negating Goliath’s advantage. They, in their resistance may be creating market pressures that magnify the bandwagon effect when a victor appears. I’ve been studying recently Andrew’s excellent paper on price discrimination. He had a really interesting reference to Scott Bradner’s list of “10 choices that were critical to the Net’s success.” So I Googled and found this version reported (and edited) by Dan Gillmore on http:// www.siliconvalley.com/mld/siliconvalley/business/columnists/4029770.htm. I found #8 possibly relevant to this particular discussion: “8) International telecommunications standards bodies reject TCP/IP, then create a separate standard called OSI. TCP/IP, remember, was designed as a low layer on top of which other applications, such as e-mail, would be created. OSI was carrier-centric, a suite of protocols that included things like e-mail. Had TCP/IP been accepted and then co-opted by the international groups and telecom companies, things we now take for granted might not have appeared, or might have been under central control. One of the fundamentals of the Net is we can create new protocols on top of IP, as Tim Berners-Lee did to create the World 150 Wide Web, says Bradner – ‘and we don’t have to have permission of the carriers to do that.’” Will SIP go the way of OSI? Maybe we should hope that SIP keeps Goliath busy long enough for David to emerge victorious. Just food for thought. Editor's Conclusion: All of which would seem to hearken back to what David Reed said at the beginning of this symposium in early December - that Skype was on the way to becoming the Wintel platform for global voice. Something very significant is happening and we have as readers a much longer immersion than intended in that. There is much chaos and turmoil out there. In looking through the darkened glass we cannot conclude anything beyond the point of view that what this discussion has described is adding an entirely new point of view to VoIP for end users and small businesses - while the impact on enterprises will not be as immediate - in the long term it is difficult to see how enterprises will remain unaffected. The COOK Report on Internet May June 2005 Interview, Discussion, and Article Highlights Click on blue page numbers below to go to those pages - no live URLs in this section Functional Aspects of Skype p. 13 COOK Report: What do you see then as the context and big picture of what is going on with Skype? Henshall: Letʼs start by considering the OS. When Skype originally launched it was as a Windows platform and there was a lot of discussion that said: “Oh Skype is just another windows application”. Folk concluded that they would never develop for other platforms because with Windows they had the whole world and therefore they didnʼt have to. Consequently their first interesting strategic move was in going multiplatform - something that for a small software company was enormous an cost. But they went on to support Mac OSX and Linux. Concurrently with this they built a PDA version for Windows compatible PDAs. Consequently we are sitting here today with Skype being available on all major platforms with the exception of Palm - if you consider Palm to be major. You also have some evidence that says they are either looking at or doing work in the Symbian area. p. 14 Henshall: Skype, as you are beginning to see, changes the way people work. Way back in 2003, I was beginning to compare Skype to Microsoft Office where the office platform was all about text and email and things like that. Now I have begun to run off of two screens. A work screen directly in front and one with Skype as an application off to one side. At the moment I have Agile Messenger on my work screen. On my left I have my Skype and email applications. This is basically a communications screen. This modifies the way one begins to work and helped me to see Skype early on as something that was creating a new communications platform in a way similar to that in which Windows created a work platform. But the things that Skype is doing all have the potential to pull users away from the old world of text and email to a newer world or real time chat, IM, voice calls - in other words integrated applications that are ultimately tied around your phone and hand set. p. 15 What happens is the same thing that happens when I put Skype on multiple machines at that point. My computer rings if it is on and if my cell phone is on, that also rings. If I have a choice, I answer the computer and take the call on VoIP. If I am actually home in my house and I have one of these combined I-mate GSM cell phones, my cell phone will know to take that call on the Wi-Fi connection and treat it as a VoIP call. COOK Report: Therefore the flexibility, independence and power of the platforms that are in the consumerʼs hands are setting up a series of relationships where the patterns of connectivity are going to happen in ways that work economically and physically for the convenience of the consumer and not the cell phone company? p. 16 COOK Report: As Rich Shockey said: “Voice is just a bucket of bits.” Henshall: Right. And the question becomes where does the bit bucket or software reside? In what type of device? This type of software is just going to infect every device in some way. The key gift this next Christmas potentially is a Wi-Fi capable phone handset that looks juts like your normal phone handset except that it is Wi-Fi capable and you have a VoIP client on the other end of it and when you want down to a local hotspot or go next door to the neighborʼs, you just take your phone with you. [snip] The question here is always-on. Always ready to talk. 3G is about always-on and so is Skype . You could say 151 that Skype is just accelerating that style of working. I can sit here in my office and connect with five other people in a conference call. We can all put ourselves on mute and leave that call running the whole day if we want. Anytime we want someoneʼs attention we can just un-mute our headset and call out to them. They call it “push-to-talk.” I can have that up right now on my computer namely a “push-to-talk” network for up to five people and continue to run another iteration of Skype, side-by-side with it, and continue to take and make calls there whenever I want. p. 18 Henshall: However let me also say that I think that it is quite plausible that the structure for IP telephony that Gorsohevsky is talking about could become the real future for communications. COOK Report: So one might well say that the message here is that while Skype so far is very successful, it certainly has some weaknesses. That from the work of Henning Schulzrinneʼs students and others we certainly have a good idea of how it operates. [See http://www. cs.columbia.edu/~library/TR-repository/ reports/reports-2004/cucs-039-04.pdf for an early attempt at a Skype protocol analysis.] Therefore, folk should not sit back and assume that Skype in its current instantiation is the be-all and end-all for peer-to-peer IP telephony? Right? Henshall: Let me answer by asking a question about what Goroshevsky is talking about. It seems that he is saying that he will give anyone who wants one an exchange and that with an exchange those people are free to go and connect up to an emerging mesh. Now hereʼs my question: What happens if Skype says that the only way to solve the enterpriseʼs problem is to give the enterprise that same capability and that you as an individual can also buy the same capability - so that rather than hav- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA ing one log on (authentication) server in Denmark, anyone can have a log on server. In this arrangement the authentication servers mesh with each other and then you make your decision as an individual as to whether you want to join all servers or whether you want to join just a few subnets. COOK Report: What you are also saying is that if the Skype people are as intelligent as we would assume them to be, they have to have resources and brain power devoted to work on these kinds of enhancements. The value of presence COOK Report: It sounded to me like you were saying that if you could bring enough people together you could create an effective counterweight against anyone who wanted to take Skype in a direction that would be bad for its users as a whole? Henshall: Itʼs a nice idea! What I have in mind is a bit like Consumer Reports or JD Powers. Unfortunately Skype is really not a marketing organization at all. I encouraged them to start their forums early on. Because of those forums they have a strong core group of beta testers out there. These people have become absolute maniacs in helping Skype develop its product. Some of them are retired, some are young kids. The testers are a crosssection of people who have a passionate interest in what it is enabling. COOK Report: Do they have a community of 3rd party developers like the iPod has acquired? Would part of your idea be to identify, coordinate and interconnect some of these entities where your role becomes one of helping the developers create viable business models for use of the Skype technology? You are the guru who knows how to build what will work in that world? Or I would come to someone like yourself if I am trying to figure out how to best apply this technology in a way that supports my on-going current business? Henshall: Yep. All this would be key elements of what can be done. I believe that we are still in the very early days of what the Skype API is and where people actually are. But I think things could accelerate very quickly - especially once the first real application that solves presence comes out. This will provide a wake up call to more than a few people. COOK Report: How do we know what to look for? Henshall: The part that I donʼt think Skype really understands is one of the most valuable: It is all about presence. While Skype offers free person-to-person calling, the larger question is what can you do when you have presence? Symposium Discussion Dec 1 – Jan 4 p. 25 December 12 Dave Hughes: Hurrah! I just KNEW that Skype would work if nothing else would! I just got a Skype call, from Tsering who was all the way over at 13,000 foot in Thame, Nepal, (12 hours difference - Sunday night here, Monday morning there) in the classroom with 9 Sherpa kids. 5 miles and three Smartbridge Wi-Fi radio relay hops from his Cyber Café base in Namche (on the Everest trekking trail) then over TWO satellite jumps to the US, (one with only 64kbps bandwidth) then over the net in the US to me! VERY clear! Then he called Mingma Sherpa in Pittsburgh from Thame, had a good clear talk. Mingma will now be able to teach those kids English, ORALLY, and not just by email, as well as use the link to instruct them in spoken Nepalese, how to better operate their classroom computer which Jim Forster of Cisco donated last year! Hey those kids are gonna get educated in spite of their remoteness! Snip David Reed: Well, I have gone on record (at the MIT CFP working group) as saying that SIP may have missed its window, because of Skype. (And I was a big fan of SIPʼs potential). 152 SIP could have been what Skype is becoming, but the SIP community has been trying to replicate the walled garden before deploying. They are destroying the value of open interoperability that was in SIP, just as Skype is opening its APIs to get the boost of third party developers. SIP should have won, because it is an open standard, but the desire to create a business model that captures the old unsustainable voice revenues of the RBOCs has seduced Cisco and its customers into waiting and making the standard more complex. Unlike the old days of the Internet, where interoperability was the centerpiece, the likelihood that a SIP phone will work with one from another vendor is near zero. There was a reason that the major IP trade show in the early days was called “Interop”! So now instead of innovating to make SIP work as simply as Skype does out of the box, the business strategy of the access providers is: attack their best customers by finding reasons to block Skype traffic. This reminds me of the IT departments who tried to keep department managers from buying Apple IIʼs because they were afraid that their budgets and power were at risk. Also reminds me the suicidal behavior of the Bluetooth consortium - (in contrast with the 802.11 vendors). Sending voice over IP is trivial. Thatʼs not the technical problem. Getting scaled adoption is hard, and a common standard that works simply was required. SIP could have been a contender. It isnʼt going be. And I think its own “proponents” killed it. This is completely analogous to what happened with Unix vs. Windows. (It is balkanization vs. a common platform). Open platforms can win, but a groupʼs self-interest in cooperation and coordination is often poorly understood by the members of the group itself. Linux, on the other hand, seems to be growing (Linux is now much larger than Macintosh, in terms of desktop market share. In server market share itʼs been The COOK Report on Internet May June 2005 dominant for longer). What makes Linux win is that the groupʼs interest in interoperability is identified and managed, as opposed to ignored and frustrated by its own members. p. 27 Reed: Skype is a completely proprietary system. I have studied the behavior of Skype as I use it, so I think I know more or less how it works in gross detail. Perhaps I can even find out more. But they donʼt follow any published standards above the TCP/IP layer. Their protocol for “just working” when behind a NAT box or firewall is elegant and simple, but it isnʼt based on standards-based NAT traversal (such as STUN or UPnP). They donʼt support SIP interoperability, though there should be no problem interconnecting to SIP if they feel they need to at a gateway. They use nonstandard presence protocols. And they exploit end-user machines, even when you arenʼt making phone calls. How do you know if they are doing bad things to your machine behind your back? Just look at the active connections on your machine using netstat or whatever your OS provides. What are those connections to Japan or India doing? Are there security risks? You donʼt know. Do you care? Maybe not. I tend to trust them, but all of their programmers are in Estonia, so what do you really know about what the code you downloaded does? They encrypt all the traffic, but what kind of key management is involved? [Snip]. Reed: They just recently defined and published some APIs by which third parties can use their protocols. These APIs let you do some rather nice things, and build things that go well beyond telephony. They are a lot easier to use than developing a whole SIP client, even if you have the open source code. Library APIs are much more productive than source code, even though not transparent. And remember, if you build on the APIs, you get a huge and growing installed base, for free. The base will support those APIs. This is pretty nice for someone who likes to build stuff for a market. Skype can control what it chooses to control. Of course they own their code - thatʼs how copyright works. They can license it on ANY terms they choose, and charge what the market will bear. They choose to let users use the binary for free, and sell Skype-Out accounts. Will they sell the later versions, or merely charge companies for the right to interconnect? They can do what they want, and will probably choose to grow their user base to create value (by Metcalfeʼs Law and Reedʼs Law valuations arising from the value of connectivity options and groupforming options that grow faster than the user base grows). Well, Skype is doing what Microsoft and Intel did in the early days of personal computers. It has created a platform that is *very* attractive to third-party developers, because of its size and ubiquity. It has invested in market share, and it is now opening interfaces and architectures that allow for others to help build value around Skype, while maintaining control of a core, and acting benevolently to those who choose to enhance the Skype platform. This is not at all like true “open source” behavior. But it is *very* attractive to both customers and partners. Remember, in the early days, it was Apple who made the mistake of not supporting its developers. They screwed their own partners, by competing with them, or making sudden changes that disrupted its partners. They deliberately harmed both peripheral developers and software developers who got too close to Appleʼs customers. On the other hand, the Unix companies (Sun, DEC, HP, IBM, ...) deliberately developed incompatible “features” that made it impossible as a third party hardware or software developer to be a Unix vendor. One had to choose which company one supported. They acted as if Microsoft was insignificant, and as if the other Unix suppliers were total enemies. I.e. they destroyed their own advantage of maturity and compatibility among themselves in the areas where compatibility and capability were advantages they “owned.” As a result, Microsoft/Intel was the truly open, binary compatible platform, and 153 until about 1988-89, they focused on building shared value with their 3rd party hardware and software partners. IBM was the first casualty, followed by Lotus, Wordperfect, Novell,... Skype can (and probably will) play this game. They need not be truly open they are open enough, and compatible enough. p. 31 Reed: I have spent my career arguing for open systems like SIP was supposed to be, not open systems like “Unix” turned out to be. I was going to write a long comment, but instead let me use a comment that I think is due to Peter Drucker (at least it reflects his way of thinking). The way to succeed in business is to pick the best customers, and delight them. And the crucial caveat - the best customers are not the ones who always buy anything you sell - those are *your* best customers, not *the* best customers. The best customers are the ones who will teach you what you should be selling. The following is how it applies here: SIPʼs vendors have defined their customers to be phone companies. Skype has defined its customers to be people who live a communications-centered life. Itʼs impossible to delight a phone company with voice over the Internet. The people who live a communications-centered life will teach you what really matters. Those people are *not* happy customers of the phone company. Itʼs still possible to beat Skype with SIP, but the current SIP vendors (such as XTen) have no clue whatsoever! To win, you have to delight some customers, not participate in an illusory “market” for “technology”. Hereʼs why Dave Hughes matters. Heʼs delighted! Iʼm sure heʼd be even more delighted with a truly open system. But he canʼt get an open system, with full interoperability, today. And he canʼt be- COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA cause the SIP people think the game is about making the incumbent phone companies happy. So openness (as defined by market-enhancing and marketdelighting interoperability and ease of use in many conditions) plays second to controlling the users. running voice and call control over port 80 (HTTP) to evade firewall restrictions or running both call control and voice on a single transport association. Thatʼs what happens when you have MBAs run your company instead of true entrepreneurs. - You can do security by obscurity. Open standards need both entrepreneurs and innovative technology. Iʼm not sure which is more important. Microsoft got where it is by entrepreneurship, and a slight bit of open APIs (partly contributed by IBMʼs PC architecture choice). X.500 was “open” but controlled by phone companies (well the OSI, which is a consortium of phone-thinking people). Internet mail headers were “open” and entrepreneur-led. It all starts with the endpoints, and delighted users trump slow-moving phone companies. When Pulver started FWD, I thought there was a chance for SIP. But when the SIP vendors didnʼt embrace it enthusiastically, that revealed FAR MORE about how the vendors think. SIP Must Interoperate While Skype Need not Do So p. 32 Schulzrinne: Not being an open standard has a few advantages: - You can limit yourself to working with only one Internet-to-PSTN “carrier” (Skype) and one, global authentication server; many of the problems experienced by SIP users are configuration problems that are much harder to solve if there are multiple service providers. I readily admit that the IETF community has not nearly paid enough attention to configuration and diagnostics issues. - You can do things that would get you in trouble with corporate security folks or any network engineering group, such as - You donʼt have to worry about interoperability as you control all the software. The success of non-open applications in general, probably has little to do with technical superiority, but it does point out the real costs of open systems in terms of testing, brittleness in the field, customer configuration and the like. The Web had it a bit easier, with essentially one or two dominant pieces of software, on both the browser side (Netscape and IE) and the server side (Apache and IIS), but even there, the experience has often been frustrating. Continued dominance of proprietary applications, as shown by IE (ActiveX), also often has little to do with technical superiority but rather, with bundling advantages. Skype and Microsoft certainly enjoyed that advantage. Neither FWD nor iptel.org nor all the other providers of similar PC-to-PC services had anywhere near the name recognition and the ability to cross-market. (I also suspect that VCs, having “learned” the lesson of earlier PC-to-PC VoIP failures, ignored the fact that the deployment of broadband made this much more viable than attempting to do voice-over-modem.) Iʼd be curious what you would cite as evidence in SIP of the sinister influence of PSTN types. Yes, there are things like “early media” that are influenced by the need to interoperate with legacy systems, but they are very much at the margins and donʼt really shape the overall system. Reed: Sinister is your word not mine. I donʼt think it is sinister. Itʼs just sad to waste time pursuing those customers. I can tell you that many of the SIP vendors are spending lots of time with those customers. And they build their business plans based on numbers from those customers. 154 p. 33 Jennings: I think a better path is to take a balanced approach of fixing the network such that real time P2P applications can work well on it while at the same time being realistic about the deployment of IPv6 and NATs. Some guy called D.P. Reed has this great paper about the “End to End” principle that I strongly believe in :-) To do this, I have been working with the NAT and Firewall vendors. For all intensive purposes there are 3 vendors of home NATs and threee vendors of non-home NATs so itʼs not that hard to reach them all. The goal is to make end-to end-possible but still allow administrators of firewalls to impose the policy they wish. If they want to stop a certain type of traffic, they always can do it so the goal is to make it easy for them to do and make it so that they donʼt accidentally kill other things in the process. Reed: Most of the network works fine already, on a performance basis. Itʼs not clear we need to add stuff to create QoS. Perhaps we should delete some stuff (like firewalls). I run real-time collaborative p2p applications (Croquet) between my home in Boston and Cary, NC; Magdeburg, Germany (former east germany); and Palo Alto. Often I use hotel networks, too. The latency and jitter is quite acceptable. What the Technology Must Do in Order to Please the User p. 35 Jennings: Iʼm all ears on this Iʼm very interested in what to improve. If standards donʼt matter, then we need to get the meetings moved from Minneapolis to somewhere nice and warm :-) You mention that people really like Skype and that it delights real users. Can you dig into that a bit and get specific - What is it about Skype that people really like? Reed: Iʼl let other users like Dave Hughes chime in. No particular order, except #1 is by far the first among equals. The COOK Report on Internet May June 2005 1. It just works out of the box in minutes, no matter where you are (zero configuration setup). Shockey: Yep. It just works. Even a certifiable SIP bigot such as myself uses it all the time. Reed: 2. The voice quality is excellent. Shockey: Nothing like the ILBC codec that is the best in the business. Now Skype did have to license that code for real money and you canʼt get it in other products like XTEN unless you pay for the licensed version of the product. http://www.globalipsound.com/ Reed: 3. Conferencing is trivial to do, and free. Shockey: Yes, but Skype had the luxury of vertically integrating all the various elements of a coherent VoIP system without having to interoperate with other implementations as SIP must. Reed: 4. No phone numbers just names, you start with “presence”, and typically use the IM feature to ask politely, with a topic in mind, if the other person wants to talk now or later. Shockey: Ditto but that is a User Interface issue that now everyone is going to mimic. Reed: 5. No “call waiting” - you can blend multiple IMs, voice calls, conferences, ... with a nice User Interface that exploits it. 6. You donʼt have to hold a telephone in your hand when you want to use a mouse and keyboard to do other stuff. Shockey: But you can do that with any decent softphone such as XTEN Reed: 7. And of course, itʼs free (or rather, since you are already paying the cost for Internet access, and Skype uses such a trivial amount of extra resources after that, itʼs a rounding error). 8. Finally (this was a bit late), Skype just works across OS X and Windows, and a few less popular platforms. In contrast, every SIP user experience design I have seen (such as XTen) is modeled after the POTS User Iinterface that evolved based on the restrictions of a central, legacy switch, with dedicated circuits, 10 key pad, special case conferencing. And every SIP install is a nightmare of “settings” and failures to operate over various firewalls, tunnels, etc. Admittedly, you could do everything that Skype does based on SIP. But to do so, youʼd have to admit that telephony as we know it sucks and is worth redoing all over again to make it fun and productive. Heck, maybe the generality of SIP might even be exploited so that you can get a voice bandwidth of 16-16KHz, and a dynamic range of 96 dB, rather than the typical 3 kHz/8 bit audio that is the best one can get from “toll quality” sound. I could play my flute for my daughter when Iʼm on the road. My point is that the *potential* for delight *is* hiding there in SIP. But the folks who do SIP have invested no time thinking about it, much less time making it happen. Stop hiring engineers and people who think writing standards is their idea of a great way to spend a weekend, and start hiring creative types who have strong views of how cool things could be. XTen, for example, isnʼt cool - itʼs just a picture of a phone on my screen. p. 39 Reed: To the extent that your blinders only lets you see the world from the perspective of “carriers” (and now Iʼm including wireless and fiber, to include Francoisʼ blinders, which are typical of the optics-heads) you miss the essence of what is happening in the world. Despite the enormous “narcissism of the operators” (who cannot see that the world of communications is NOT about them), the real world of communications is about the messages, not the bits, and the users, not those who would claim to own them. COOK Report: We had a similar argument about 6 weeks ago. Is there any chance that we could agree that - like it 155 or not - for the foreseeable future (several years?) SOMEONE besides just the end users will have to operate networks and provide wide area service? Eventually mesh nets and viral networks may combine with dynamic optical technology to push standard operators into the so-called dustbin of history. But for some number of years this is not likely. In my opinion at least. A Changing Role for Operators p. 39 Reed: Of course operators will exist and operate networks. That isnʼt my point about the “narcissism of the operators.” The narcissism is the idea that they deserve to play the role of defining what the services will be that are supported by the meaning-free bits that traverse their networks, among interoperable endpoints. Francois is committed to putting in pipes to carry those bits. I suspect he would like his pipes to be the preferred solution for as many bits in as many circumstances as possible. More power to him, but my point is that other than doing a good job on those bits, he has no leverage to force certain kinds of bits to traverse his pipes. Bits that make up video streams may travel over his pipes or over wireless connections or via satellites or over packets delivered using Bit Torrent asynchronously. The users now have the leverage created by interoperability, and they are using it. They do “route around” attempts to tie services to particular sets of pipes. Thereʼs a fantasy out there that fiber is so wonderful that the first fiber to a home will “own” that home forever, and allow lock in. I donʼt buy that - itʼs just another way of saying that no one will ever need more than 300 baud because a person canʼt read that fast. Shockey: Well if you put 100 channels of High Definition NFL football on it you just might. :-) Reed: If the operators get over their narcissism that the world depends on them, COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA so they should be given all kinds of rights and privileges, they have perfectly fine businesses to run. Iʼm sick and tired of the wailing and crying and poor mouthing I hear from operators. The sky isnʼt falling. Shockey: Well, the sky is falling if you are a LEC and look at their gross revenues and margins on the landline sides of their business. Its not a pretty sight when you then factor into your business model that you will lose 1/3 of your residential business to Cable Operators within 3-4 years and perhaps 30% of your enterprise business to SIP trunking at the edge and IP Centrex from the IXCʼs -- if they can hold on long enough. Reed: Itʼs not the operators that will define the services, or “permit” them to exist. They will get paid, to the extent that a lack of capacity will find entrepreneurs willing to fill it. Itʼs just that the current operators need to realize that they havenʼt got a lock on that new business, and to get started on figuring out how to remain competitive. Shockey: David, itʼs just that traditional telephony carriers do not want to be turned into gas, water, or electricity utilities where they have to live on reduced but highly regulated rates of return much lower than they are accustomed to though I need to research this more. The ILECS still think they are high tech businesses when they are not... they are bit pushers. They are totally enamored of “new services” or “content delivery” where we both know that those services, as Odlyzko has proven, are not all that profitable and you know can be delivered at the edge by anyone. Reed: Else they are going to be the next Kodak or Polaroid - companies who thought the film stock and patents they held were the core of what photography was going to mean to its users forever. Now itʼs about ink, image sensors, metadata, coding, and digital storage. Shockey: Well, I still think there is a good business being a bit pusher. The ultimate economic question I have is since we can now conclusively prove that Voice is simply a edge application on the network (the Skype demonstration), which means its real marginal value is zero (like email). What will happen to the telecom industry when you yank over 200B in revenue right out from under it? OK ..50 Billion. Have you looked at the bond ratings on Qwest recently? Symposium Jan 5 – 15 A Balkanization of Personal and Enterprise Communication Trends p. 41 Coluccio: The discussion that has been about losing the telephone number in favor of a Skype tag or other “personalized ID” I find to be somewhat of a distraction, while also a bit amusing, because it once again demonstrates how the discussion on this list has a tendency to balkanize the telecom universe, and ultimately focusing primarily on “personal” communications trends, as opposed to how those trends will eventually meld with the communications traits exhibited by large commercial and government service characteristics. Yet, wide-sweeping projections are made that appear to encompass all of the above, when they are, in fact true, but they are true most likely for only the most personal level of individual usersʼ communications needs and provisioning, even if they find their way into enterprises. Like a PC is personal to each user, e.g. Skype-like applications, I firmly believe, will come to permeate every aspect of personal and office communications, but those applications will only “displace” the traditional attributes of earlier services in a limited percentage of total point solutions. Where the newer Skypelikes will dominate most demonstrably, in my opinion, will be as incremental personal applications. Just as email and IM have grown to become, today. That is a far cry, however, from putting telephone numbers in their graves. The world as we know it is not made up of IP coders and IETF delegates, despite the size of this list becoming what it 156 has, and despite our losing focus of the larger flows that take place each day that have absolutely nothing to do with the open Internet, and never get counted on anyoneʼs stats because they take place behind closed optics. Multiple VoIP Markets p. 42 On December 30 Schulzrinne: Having read some of the discussion, it seems weʼre simultaneously talking about three different things: - product design - wideband codecs - protocol design Only the last one has anything to do with SIP. It is pretty clear that many SIP implementations have fallen short on the configuration end, with lots of configuration options that shouldnʼt be necessary to be exposed at all to normal users and inconsistent labeling of the two items that are really necessary. A good SIP device should need exactly the userʼs email address/phone number (for corporate) or carrier SIP URI (for FWD and the like) and a password. Making software look like cell phones or office phones is part of the same User Interface disease that makes certain vendors, even ones with a tradition of UI excellence, convert their media players into jukeboxes and DVD playback software into silvery boxes with 7-segment green play time counters. The whole notion of “skinning” is beyond me; strangely, Microsoftʼs Windows Messenger SIP tool seems to have largely escaped it. As far as I know, it predates Skype. I guess early automobile designers couldnʼt resist making their horseless carriages look like, well, horseless carriages. I should point out that the Pingtel phone, a SIP phone device, was designed with wideband codecs in mind, although Iʼm not sure how widely they got out of beta with that feature. The COOK Report on Internet May June 2005 snip Schulzrinne: We clearly have at least two different VoIP markets: first-line landline replacement (Vonage, AT&T and the like) and the talk-cheap-internationally-by-PC market. User feature requirements for the former seem rather different than for the latter. p. 47 January 11 COOK Report: I have added Raj Sharma of NexTone to the list. Rajʼs principal interests lie in the harmonization of protocols for the IXCs, ITSPs, ISPs and ILECs, such as SIP and H.323 along with a long list of proprietary ones made by all of the VoIP fieldʼs popular vendors. Reed: Cool. Welcome Raj to the debating society! But to get down to brass tacks, and semantics that show a broken mindset, “harmonization” (in its typical telecom meaning) is not going to be a winning strategy. Harmonization is what a bunch of technocrats, relying on a government enforced oligopoly (i.e. the ITU) thinks it has the power to do. I.e. shades of the 3GPP, or NANP. VoIP is inherently happening at the edges (whether they are PBXes, SIP phones, etc. at the hardware level, and Skype, FWD, AIM, PTT, Blackberry email as the viral integrating technologies). Viral Communications Stastny: To which I fully agree. By the way, David, congratulations to the “Viral” paper. Since it is a draft and quite old, does an updated version exist? Reed: Richard - the viral paper was written for a special issue of the BT Technology Journal (October 2004). That journal is now out, and I commend it to you... http://dl.media.mit.edu/viral/viral.pdf may be the same version you have - I think we should probably replace it with the prettier published version from the BT Journal, now that you point it out and assuming we can get a PDF of that. Editorʼs Note: The special issue of the BT Technology Journal is on the Media lab Web site and is well worth reading. However I steadfastly disagree with David in that I find the general sweep of the May 19 2003 draft much better than the two early articles in the BT journal that are intended to replace it. know you and your ideas have turned up absolutely unaware of this paper. The ideas are certainly out there. But the May 2003 version of the paper wonderfully ties them together in a FRAMEWORK or a pair of lens through which to see the world. We need frameworks I think. And this is a REALLY REALY GOOD one. Reed: You may also be interested in our group website: http://dl.media.mit. edu/viral/ and the cross-MIT sponsored program Andy and I have launched with Dave Clark and Charlie Fine called the Communications Futures Program, which focuses on the evolution of the architecture of the communications industry (website under re-construction). Now you mentioned special issue of the BT Technology Journal (October 2004). With that as a lead, I found in Google And the broader embedding of the viral work in human networks, our bigger vision called “Organic Networks” in that special issue: http://dl.media.mit. edu/BT-vco.pdf Donʼt want to be seen as hawking CFP or the Media Lab here, though. COOK Report: David, Iʼd like to do that for you. :-) Seriously I am amazed - People on this list are gradually beginning to read it and uniformly say WOW. I advertised and endorsed it to a private list of David Isenbergʼs that some of you are on there again not much reaction. Last night I sent the PDF to Sebastian Hassenger and pleaded with him to read it. Talked with him today. He has read it and remarked that he finds it really wonderful and wondered how he ever missed it. Sebastian is Senior strategist for pervasive computing in one of IBMs WestchesterCcounty, NY Labs. He was on one of my lists in the August October 2003 time frame. To the extent that I understand what he does it is to survey where all this stuff is going and make sure his colleagues at IBM understand. I told him what we are doing and he agreed it was right down his alleyway and agreed to join. David, there seems to me there is a theme going here... many, many people who 157 http://www.media.mit.edu/publications/ bttj/ Competition and Wireless Paths WiMAX Troubles p. 52 WiMAXʼs fate (yet undecided, though I think it will fall into a miniscule shadow of what it could have been, just like Bluetooth has) proves nothing - thatʼs the mistake of synecdoche, i.e. the case where a metaphor conflates an instance with one of its categories. Shockey: Iʼm increasingly interested in this. I have heard anecdotal evidence that WiMAX has some big problems with fog and other extreme forms of weather that is peculiar to the RF absorption characteristics of the spectrum WiMAX typically uses. Has anyone heard anything more about this? Reed: WiMAX has bigger troubles. Itʼs no different than cellular from the 50,000 foot perspective. Itʼs not opening a new market, but merely using the same technology to try to start a service that competes with a universally present service. The only advantage it might have is that it is not being positioned as a mobile service with roaming, so it can be rolled out incrementally. But cellular data services are easy to add. Somebody should study Christensenʼs book again, and notice that wireless broadband need not enter the market in the markets that have already been cherry picked. 802.11 is a much better COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA starting point - every computer user already HAS one endpoint of that system, and will buy whatever feeds their itch. 802.11 reaches the telephone poles in your neighborhood, if you put a bridge in your window. And if there are no telephone poles, it reaches your neighborʼs apartment or your neighbor who has a fiber with lots of spare capacity for now. Thatʼs HFW (hybrid fiber(cable)wireless). A lot cheaper and more upgradeable than APON, EPON and other PONzi schemes. Cellular is already taking share from some wireline services at a rate that erodes the profit pool. There are lots of other wireless paths beginning to unfold. Cable competition (facilities based competition in HFC) is finally born in some regions, despite the fundamental socialist intuition that no economy can sustain 2 fibers to any home (a lot of hooey based on the old Big Lie that demand is inelastic for communications). band audio codecs is that he says for the future it will be like the difference between black and white TV and color. This is where he sees services differentiating themselves once price is no longer an issue. The thinking seems to be that market share will be determined by whatever platform delivers the best sensory experience. This is an area where Skype already is quite far ahead. COOK Report: Certainly visual presence canʼt be far behind? All manner of web cam devices are becoming very cheap.Therefore, it probably will not be long at all until Skype or its competitors do video as well as audio? As monitors get bigger there will certainly be room for a video window. As you move all kinds of presence to the edge on an individual basis, this should drive increased demand for bandwidth and for filling up fiber. The audio aspect is open to improvements in interesting ways. I have seen reports from Stuart Henshall that he has heard that Skype is working on spatial positioning of voices in a conference call. There is always scope for something like haptics where you can project some element of a personʼs physical presence such as a heartbeat to come into play here. In the grandparentsʼ scenario, the sense of being there and being able to listen to the kids play as opposed to having five minutes on the phone with them is worth thinking about. This of course becomes possible if the technology works and as long as the service is not metered. Enck: Yes - all of what you suggest is absolutely true. I think what Duric is saying is indicative of a growing realization that use of peer-to-peer audio communication softphones or software like Sype is just fundamentally different than products based on SIP that must use ordinary hardware telephones and be able to connect effortlessly to the PSTN. You have on the one hand what he is saying and on the other hand rather negative responses coming from the SIP folk and the SIP --using hardware community. This is indicative of the Skypeʼs having driven a wedge down not only the proprietary versus open standards issue but about what the very nature of the service is. Zennstrom has never made any bones about the fact that Skype was originally basing its interface and user experience on instant messaging adding in effect instant voice to instant messaging. They were never trying to recreate the telephone experience. Alan Duric is the CTO of Telio, not the one that every one is talking about in the States but a company that is basically a Norwegian Vonage. Duric used to work at Global IP Sound where he was one of the authors of the Sype audio codec. His description of some of the things that are in the development pipeline with broad- COOK Report: This gets back to what David Reed was saying last December on the mail list about Skype not being an IP replacement for the telephone but being designed for strongest appeal to those living a communications centered life. One has to really re-orient oneʼs mindset in approaching this. James Enck Interview p. 57 Future Possibilities: Improvements in Presence and in Audio 158 Enterprise Concerns about Skype Encourage Competitive Products p. 58 COOK Report: Perhaps this is because of the very negative reaction that Skype gets from enterprise security people? Would you describe what you hear from enterprises as to their use or non use of Skype? Enck: I wrote a piece not long ago on Accenture. It turns out that this piece flushed out two other guys. They have project teams working in Pakistan, and other teams in Amsterdam, Madrid, Vienna and London. First of all within Accenture itself I have confirmation that top managers and even board members are aware that this is going on. While there is no official policy pro or con, they have never said to their employees that you cannot use Skype. Apparently at least one senior board member of Accenture is a Skype user himself and does so within the business. Now when they are out at client sites what they do apparently is set up their own free-standing Internet access point. They would work through the local telco to construct a VPN that would enable them to set up their own access node independent of the clientʼs network. Or perhaps they would just bring in an independent DSL line. Therefore, the client network is never at risk. It is Accenture within Accenture. The reading I was given is that they are currently interested enough in it and want to see where it will take them such that there is no effort made to stop it or restrict it in anyway. The folks I have talked with stress that, while they are not going in clandestinely and using this behind a clientʼs firewall, it is something they feel comfortable enough to use within their own organization and that they are not discouraged in anyway from doing this. COOK Report: What do you hear about the dangers and risks of a peer-to-peer system like Skype? Enck: While I am not a security expert, I The COOK Report on Internet May June 2005 have certainly noted Melissaʼs concerns about an open Port 80. I have also noted the opinion of Dimitry Goroshevsky at Popular Telephony http://www.peerio. com/ and the folks at Nimcat Networks in Canada http://www.nimcatnetworks. com/ that any IT manager who turns loose an application that can punch a hole in a corporate firewall needs his head examined and that is why they have built into their application an administrator role where central control is required to switch functions on and off. COOK Report: So isnʼt this another way of saying that what is going on here will probably evolve into new software packages, new devices and new companies? Right? Enck: Absolutely and I think that there may be future iterations of Skype that embrace enterprise security concerns. It canʼt be something that is perceived ass a renegade product. They will realize that if their product is to find widespread enterprise use, it has to be controllable from the IT managerʼs view point or it will be seen as an unacceptable threat no matter what the perception of benefits by the other folk in the enterprise. Popular Telephony p. 59 Enck: Over the past two or three months their press releases have made it clear that they have a working relationship with Texas Instruments (TI). COOK Report: And TI would have denied Goroshevskyʼs assertion were it false? Enck: I think that indeed you can count on that. It looks now that TI has ported his software onto its DSP. Given that TI is there I would be surprised if Broadcom is not also involved. My surmise is based on the fact that Broadcom is a financial investor in Nimcat Networks which is Goroshevskyʼs principal competitor. Also I know someone in the product and development unit of a large European electronics firm. These folk have actually done trials of the Peerio application in their laboratory and had good things to say about it. COOK Report: The software resides as firmware in a TI Digital Signal Processor (DSP) chip then? Where would one find this chipset? Enck: It would be in the desktop VoIP or hybrid kind of phones that are coming out. The ones that they, in late February, announced from Grandi (a Chinese firm) is a good example. That phone has a PSTN connection, two Ethernet jacks and also a SIM Card slot. Popular Telephony has signed agreements with about 6 different makers at this point. They have Gateway people involved. COOK Report: Their software loads into the DSP as firmware? Enck: Yes. It is firmware. Middleware or whatever you want to call it. Theoretically at least it can sit on mobile handsets or on virtually any other device you want to think about. COOK Report: How would it work? Enck: I think it would be conceptually similar to some of the clients developed around Skype by third parties. For example there is a product called the DualPhone, a DECT phone for Skype <http:// fa86dd8e8eff5070c1256f1c0040dee5. dualphone.net/>. It is a cordless phone that is available in Europe. It is both a wireless PSTN phone and one that affords you a connection to your PC and to your Skype contacts. You can access both and choose what kind of call you want to make from this single handset. Conceptually that is how it would function. What the mechanics at this point are going to be I am not sure. Whether there is to be a purple button with a “p” on it to launch Peerio for example? Or whether it is in the numbering or in how your address book is structured? COOK Report: But presumably it could also be distributed as a software client? I understand that like Skype it is peerto-peer in its structure but presumably it is how the “peer-to-peerness” is implemented that is different? 159 Enck: Yes. I think that is a good example of the difference. Peerio is a bit of a black box. I had a conversation with Bram Cohen, the guy who wrote Bit Torrent. Although he was not aware of any of the P2P voice systems I gave him a description of what Dmitry was doing and said that I had asked Dmitry whether Peerio was similar to any other P2P application? Dmitry had replied well, yes and no. Conceptually it was similar to something called Chord. Later I read some academic papers describing Chord but did not understand them very well. However when I mentioned Chord to Bram Cohen, he immediately said that he understood what Goroshevsky was doing. For this I conclude that within P2P coding circles, this is an architecture that is actually understood. Skypeʼs Cellphone and Wi-fi Direction p. 60 COOK Report: How would you describe the wireless wi-fi direction in which Skype appears to be headed? I think Stuart in his conversation with me was suggesting that Skype viewed its value more in mobile technology than in convergence with the PSTN. Enck: I think that is correct. Certainly in Europe the mobile area is where a lot of the arbitrage opportunity for subscribers exists. The reason for this is that if you are traveling around Europe, you pay very heavy charges for roaming. This is an area that costs businesses a lot of money and where they would be very keen to make savings. Based on anecdotal evidence I get people from the States seem to be happy to go into a T-Mobile or BT commercial hotspot and pay seven pounds for an hour because in that hour they can make forty dollars worth of international calls over Skype. Very worthy of attention in this area is the announcement yesterday (March 3) regarding Broadreach. They are huge in the UK. They cover all of the London train terminals, as well as all the MoTo roadside truck and car stops that are really somewhat like miniature shopping centers. They claim that 500 million people is the annual total of the daily visitors COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA to all their locations, although obviously people arenʼt going to be carrying their laptops on each and every visit. Broadreach is basically a wi-fi managed services company that is hired by the railroads to wire up the train stations for wi-fi and so on. They actually do give services to a chain of coffee shops called “EAT”. They manage and connect to the Internet the networks of the various enterprises that buy their white label services. They seem to specialize in white label wi-fi networks for retail and travel organizations. One of the companies I talked to recently that does work in this area was getting about 350 to 400 pounds per month per site. They have a little Linksys wi-fi node and DSL connection and come by and check it out every few days. COOK Report: But a London train station takes much more substantial equipment? Enck: Oh yes. You can bet on that. Meanwhile some of the locations are quite desirable and this will expand enormously the Skype friendly locations for people to work from. The release was worded very strangely and I am not precisely sure how the use arrangement will work. p. 60 COOK Report: How far are we from these dual use handsets being on the market? Two or three months? Enck: I think that is a about right for the enterprise market. And perhaps another two to four months beyond that for the general consumer market. My reading of the handset makers is that they are scared. They have been the victim of the carriers for too long. They have seen the carriers hammering them down for too long. They are begining to see that the carriers have a limited shelf life so they have to begin to ask whether they want to continue to be a slave to this dying industry? Or do they want to get in bed with an application developer like Skype. You can see it now not only with Skype and Motorola, but you can also see it with Microsoft Nokia announce- ment that shocker everyone so much a few weeks back about allowing direct connectivity between Windows XP and Symbian based handset in order to share music. This will immediately eliminate a certain amount of carrier revenue. Who is going to pay a dollar fifty for a song download over the network when you can simply down load to your I-tunes and the upload it to your phone? Nokia signed an agreement with Macromedia developers to allow more independent applications coming in. They have also opened up a Python forum. Python is apparently well tailored to writing things like SIP applications. Bit torrent is written in Python and the fact that they are opening up the Nokia platform to this kind of programming language that is associated with wild disruptive open source technology is a further sign that the handset makers donʼt want to be enslaved to the carriers. I think Motorola was really sticking up its middle finger at its traditional customer base and saying “look we want to carve out a sustainable stake in this sort of value chain because we think your traditional model may be in trouble and we donʼt want to be on the wrong end of that.” I think also that the Motorola guys will have seen the Taiwanese makers HTC, or High Tech Corporation trying to court Skype. They will have also seen the Skype I-Mate announcement and find themselves getting very paranoid. COOK Report: Are you seeing anything from Skype about how soon they are going to be out on Symbian? [Editorʼs Note - Global IP sound has announced a version of their Voice Engine codec for Symbian during the week of March 7.] Enck: No and that is an important question. If they really want to cover the mobile front, they do need to get it out there. pp. 62-63 COOK Report: What we are talking about then is a really well entrenched peer-to-peer system. The ability to stop or root out such a system as long as access to the Internet is not licensed and controlled along the lines 160 described by the author of the Digital Imprimatur is generally nonexistent? To try to control such networks would render so many other things inoperable as to effectively destroy the Internet? Enck: Yes. I donʼt think anyone can root it out. It is here to stay. That is one of the things I like about it. There is no reasonable way that you could gain control of what it does. If you look at industry trying to create alternative products that will do what peer-to-peer networks do and create what people like about them, then they have a really lousy track record. In my view practically everything that has happened in Internet/telecom product development has been an accident. Email. SMS. These all found uses that were not intended by the developers and they have subsequently become massively transformative consumer technologies. I think instant messaging has been another case in point. What has Yahoo, MSN and AOL accomplished? What was that all about? What was the master plan behind it? It generates zero revenue for the telcos. It has caused all sorts of problems in enterprises. But they have also found ways to use it for their own ends by putting it safely inside of sub-netted playpens blocked off from undesirables. COOK Report: What are your views on Skypeʼs market place position? Do they have such a head start that no one else can overtake them? Skypeʼs Market Place Position and Prospects for More Controllable Competitors Enck - Being bigger than all the other VoIP implementations in the world combined it is looking pretty formidable. Yahoo Japan would be the closest in size. But what have they got? Five million? And yahoo is an access-based service so you are stuck with an IP phone in your house. It simply doesnʼt have the same feature set as Skype. The COOK Report on Internet May June 2005 COOK Report: What you mean is that it is rather like Vonage or Lingo? Enck: Yes. Exactly. Then you get down to the next level with firms like Vonage and Iliad in France. Time Warner Cable and so on. At this level we are still talking about only a few hundred thousand users. You put all of these together and you still get a number that is smaller than the registered user base of Skype. We are talking about over 2 million concurrent users at any give time these days and that is a helluva lot of people. COOK Report: What I am beginning to realize is that the feature set of things that can be done with a P2P instant voice program like Skype are by no means exhausted. But still Skype seems so well embedded that the competitionʼs feature set would have to be really stunning to get people to switch? Enck: Quite true. Likely the only way would be to take an entirely different approach. There is a company called Voipster in the Netherlands I have been tracking that had gone rather quiet. But I reestablished contact with them a couple of days ago. www.voipster.com/ They have a similar background to Skype in the sense that they are very small and have developer in Estonia. They were using the Global IP Sound codec but they dropped that in favor of something that uses less bandwidth because they are trying to get a start in some emerging markets. Their goal is to be an invisible partner to carriers or ISPs. Or potentially to people in the online world whether that is retailing or services. They should becoming out soon with exciting announcements. Their architecture is very similar to Skype but they allow a level of control at the administrator level with the set up of their systems that will be a relief to enterprise security people. COOK Report: In this sense they might be the software that gives Google working telephone icons for parts of its emerging local service activity? Services like this that could be used and administered by third parties rather than just out there like Skype might be more interesting to Google? And in the third world are if you can get by on less bandwidth, you have a lot of possibilities. packages may be putting themselves at financial risk if their PCs happen to be supernodes on the Skype network. p. 64 COOK Report: Kees Neggers with Surfnet6 and Gigaport in the Netherlands has certainly prepared the groundwork for this expansion. But the ideology here in the United States that says anything done by a government is, by definition incompetent, is doing us considerable harm. The direction of the technology is pushing broadband more and more in the direction of becoming a central economic utility - as vital to business activity as roads, water, sewers and electricity. Perhaps with more Skype like applications and the new things underway at Google, it will begin to dawn on people that broadband is more than just faster down load of email and web pages? I went to broadband only because I wanted long distance telephone arbitrage with Vonage but now that I am there it makes me see things in entirely new ways. However in places where cable TV is making inroads, the cable companies are saying to consumers we will double you band width speed or double the cap for the same prices as the telco. This is hardly satisfactory. I see it as a feeble attempt to circle the wagons. You could really smell the fear in the Netherlands the other day when KPN announced job cuts of 8,000 positions over the next five years. Enck: Agreed. I think what you see now in Europe is that the carriers see broadband as a new revenue opportunity with the ability to charge for overage on monthly usage caps. What they are doing is putting a usage cap on their prices. The minimum BT product at the moment is capped at one gigabyte - but for heavenʼs sake that is a less than a single feature length film. What is that in music terms? Four hundred songs? Itʼs a joke. What they want to do is sell you incremental bandwidth for a charge. Telenor which is one of the most forward looking companies in Europe tried this idea two years ago and quickly scrapped it because of overwhelmingly negative reaction. But all the telcos in Europe are looking at capped products. They sell you cheaply a basic connection with a cap or you can upgrade to the so-called unlimited plan. Unfortunately people then find that those have limits of 20 to 30 gigabytes. (http://www.bt.com/broadband/bb_info. jsp?targetSection=packages) An aside: Martin Geddes made an interesting post the other day about how people who are on bandwidth capped 161 pp. 64-65 COOK Report: How would you sum things up at this point? Enck: It is a ball of confusion! I think Skype has been an inspiration to watch. I was lucky enough to stumble on it a week or so after they launched and started to write about it. I think some people here thought I was crazy but some other clients clearly grasped the meaning of this from day one. We are talking here about institutional investors who have peopleʼs pension money at stake. I think that everything we see from them and everything being thrown at Skype by want-to-be competitors is evidence that it is a genuine force with validity and staying power. The validity is best described along the grounds of saying that users simply love the product. You plug it in and it just works. It seems to be engendering a different kind of behavior and usage from people than does traditional telephony. I think at this point if you are out there trying to sell a VoIP product on the basis of it being simply cheaper PSTN quality voice that you are on the wrong track. Symposium Jan 15 – Feb 8 p. 66 Raj Sharma: It may be useful to think of IP telephony in the context of peering: (i) carrier to carrier peering, (ii) carrier to enterprise peering, COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA (iii) carrier to consumer peering. Even in case of Skype, there is VoIP peering, specifically with its SkypeOut service, where Skype hands off traffic to another VoIP carrier to terminate the call to a traditional “ʻblack” phone. But letʼs say that all 6 billion people on the planet are using VoIP and they have all discarded their traditional “black” phones, does the need for peering go away? I will submit to you that peering is required no matter what - it is required every time two networks which have different ownerships connect with each other. Snip Forster: I think an implicit assumption in what you are saying is that the business models can support the expense of negotiating and executing these agreements and deploying equipment required to support these agreements, whereas the alternative is to design the infrastructure and end points such that the need for these is minimized. In the past the cost of all this was small compared to the rest of the business, but IP telephony can drive down the costs of operation enormously and itʼs especially noticeable as the voice service has been largely separated from the access network service -- a typical subscriber pays $X for broadband Internet access, and then $Y for a voice service, and Y<X. In some cases with certain restrictions Y=0, but in any case the separation of access from voice service highlights the competition for the voice service. Weʼve argued about various aspects of Skype, but I think no one will argue that what theyʼve built is incredibly efficient -- very, very few people required per thousand customers. It just operates, with basically no operations staff. So the challenge for these peering functions is to justify their expense. Schulzrinne: The question is whether VoIP requires special application-specific peering, beyond the already-occurring IP-level peering at BGP level. Nobody is arguing with the need for the latter. Iʼm certainly not enamored with VoIP-spe- cific peering - we donʼt do email peering or web peering, either. COOK Report: I think Nextone has a window of opportunity - (space of time before the iceberg closes in). The Question is whether the window is six months or two to three years? Viral Radio Pp, 68-69 Reed: Jere - the truly simple and cheap viral radio network is a research agenda, not a business plan, so Iʼm very hesitant to predict deliverables. Itʼs important not to confuse a clear view with a short distance. At the Media Lab, we are inventing ways to build systems that scale by cooperative adaptation to the propagation environment, that need no infrastructure. So what have we invented? Well, Iʼve been involved in filing 3 patents so far on the viral radio ideas, one on an invention done before I joined the Media Lab, and two after, and a couple more are being prepared. Not trying to be coy, but I canʼt disclose the details because of the way the Media Lab sponsor contracts work - sponsors get first looks and first chance to negotiate rights. (Itʼs a good deal to join a Media Lab consortium or the Communications Futures Program!) The work involves both techniques based on advanced software-defined radios that have highly adaptable frontends and fancy DSPs, and techniques that can be applied with mass-produced cheap radios of the kind that cost $10 per unit today - the former could be just as cheap, but the volumes havenʼt been there to drive the learning curves, though the new GnuRadio Universal Software Radio Peripheral is the next step in cost reduction on that paradigm (first radios were $20,000 per unit, USRPs under $500 per unit, ...). We also have a small NSF contract, exploring the use of networks of software radios to measure and adapt repeaters to indoor and dense urban propagation as it changes. 162 A couple of students have written and published papers describing different approaches for cooperative physical layer repeating. Another student has been developing the concepts and techniques associated with “viral broadband” - to demonstrate his hypothesis that cross-connects within towns and neighborhoods and distribution of content into cheap local storage provided by the users provides a richer set of services that includes all “broadband”, but in a much more incremental and organic economic process than traditional command and control central service providers. But as I said, this is all early stuff. We arenʼt the only people exploring the technologies that will lead to “viral” infrastructures. But I think our long-term vision is more clear - most of the other folks are trying to adapt ideas like multiantenna systems, MIMO, ... into existing legacy networks. Nothing wrong with that, in fact itʼs hugely synergistic. Google wants dark fiber pp. 72-73 Coluccio: With respect to the prospect of a company like Googleʼs building their own infrastructure and taking a slice of IP traffic off the open Internet along with it, Jere Retzer wrote: “I find that hard to swallow. Google benefits as much or more than any company from a full interconnected network. Just because they are thinking about acquiring their own high-speed links in no way implies they would remove access to their content and services.” Gill: I agree. Coluccio: True, but that is not what I was getting at. Google itself or any other like it would always have a desire to extend their content and services to users, whether those users are large enterprises or consumers, but such does not *always* require what has now become regarded as access through oneon-one connections with them. Take for The COOK Report on Internet May June 2005 example how large firms now access carrier hotels and collocation sites on their own, by virtue of lambda directs they execute through their own dark fibers, which they rent from fibercos or build on their own. Can you envisage a Google cutting deals with the Merrill Lynches and GMs of the world, heck, with any enterprise for that matter, whereby Layer 1/2 piping to their data centers and communications center avails to end users the same content and services as those companiesʼ users would otherwise have to search out over the Web? Think of it as a drop shipment in bulk capacity form (daily uploads to client caches?), if you will. A collateral benefit of this could even be seen in the heightening of security, too, as well as offloading of traffic from main access router ports to the Web, where business as usual is conducted. Gill: The problem is that ports cost money. For a few tens of critical suppliers, sure. However, it doesnʼt scale to several thousand such. Building and maintaining a fiber infrastructure that homeruns to the customer base costs money. You want to get the benefit of amortizing your OPEX and CAPEX across as many connections as possible over the same pipe. Coluccio: Looking forward, I believe weʼll see more of this type of WWW bypass situation unfolding where it makes sense, which, if not done identically to what Iʼve just described above, then in some similar form. Such capabilities will be enabled cost effectively through the inherent abilities and economics afforded by optical switching, when done for the purpose of creating quasi-permanent virtual links to SPs at Layer 1/2 from colocated network elements owned by end users in the same or colos, or even between virtual colocation centers. Gill: This looks good on paper. Places where rubber meets the road, not so much. With the rate of current trouble tickets on our pipes, there will be a full time department just managing and troubleshooting the network. This is not Googleʼs core functionality. Coluccio: There is always the other way out by using a gateway built for the purpose, as an alternative that would allow the accessing of the WWW directly, and then crawling over to the targeted Google or cache site, as is normally done. And while I view the eventualities I cited above as inevitable, Iʼm only stating them as my interpretation of a trajectory thatʼs already in motion, and not from an ideological slant. I hope that that was a sufficient amount of CO2 to douse any nascent flames that might have been smoldering. ;) And if Google currently operates using several languages now for the purpose of mass customization, and seeks through further market targeting to entice additional business through a narrower approach to customized products, then think what having a captive usership on the other end of a dedicated pipe might portend for their marketing potential down the line, sans the overhead of much other peoplesʼ services and gunk. Gill: Or license their software and run it in the customer datacenter, instead of providing pipe. Cohen: Ooh. I forgot one thing. The guy in charge of infrastructure to some extent at Google was at Yahoo before that. He had plans there to build Yahoo a global network of some sort and probably took that plan with him. Gill: Precisely. There isnʼt a quantum jump between “100% transit in one location” to “global backbone.” At some point, it makes more sense to gradually build out the backbone like a large ISP and run that between your major data centers, and to peering locations. Its a purely numbers play. AOL did something very similar to this, gradually moving from a few transit points to a global backbone with no transit. St. Arnaud: I suspect the reason that Google may want dark fiber is the same economic pressures that are driving many media companies to reduce Internet transit fees by locating services at major carrier hotels around the country. 163 Google VoIP p.74 Stastny: Another idea why Google may want dark fiber is on Tom Keatings Blog http://blog.tmcnet.com/blog/tom-keating/VoIP/VoIP-blog/google-VoIP.asp Schulzrinne: Calling this idle speculation would give it too much credibility. Unless they plan to supply their yellow page business customers with fiber or run fiber to residences, none of these VoIP calls would ever touch their network. Theyʼd only make money if they became another run-of-the-mill Tier-1 ISP, of which there is not exactly a shortage. Some of these have gone through the Chapter 11 rinse cycle, so their former stock holders (and now wallpaper holders) have paid for their capital expenses. Stastny: Of course Henning is correct that VoIP calls from the end-user clicking on the link, but on one hand they may want to connect the other side. On the other hand the idea of Google as ISP may not be too far-fetched: Think about YahooBB in Japan and Fastweb in Italy (providing FTTH in all bigger Italian cities between 50 to 100 Euro/ month). The 100 Euro provides you with at least 10MB and a set-top box containing a video-cam for real-time video communications from home. Italians like to participate in TV-shows interactively Wetzel: Thatʼs funny, instead of having access providers that desperately want to provide content with the hope of making money with it, to compensate the loss they experience in selling low cost accesses, we may see content providers which will provide accesses to end users and which will make money with both. Coluccio: It sounds similar to, but not exactly, the evolution of the cable TV cartel. Oops... did I write ʻcartelʼ? Of course, I meant to write “industry.” COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Stastny: Tom Keating was right: see: http://www.timesonline.co.uk/article/0,,2-1454225,00.html Citation: Mr Hewitt said that a Google telephone service could be made to link with the Google search engine, which already conducts half of all internet inquiries made around the world. A surfer looking for a clothes retailer could simply find the web site and click on the screen to speak to the shop. See the following quote: Google gears up for a free-phone challenge to BT by Elizabeth Judge, Telecoms Correspondent http://images.thetimes.co.uk/images/trans.gif GOOGLE revolutionized the internet. Now it is hoping to do the same with our phones. The company behind the US-based Internet search engine looks set to launch a free telephone service that links users via a broadband Internet connection using a headset and home computer. p. 78 Coluccio: Itʼs odd that you should mention Sprint-Nextel here, because AT&T is in the process, as we speak, of becoming a Mobile Virtual Network Operator (MVNO) taking partitions off of Sprintʼs wireless platforms, effectively becoming a reseller of Sprintʼs wireless services. When Seven-Eleven does this they are called a MVNO. And in my book, that is exactly what AT&T has committed to doing, as well. That sounds to me like theyʼre still trying to garner revenues from voice minutes. Doesnʼt it sound that way to you? The company has always been good for making a splash far in advance of perceived threats. As when, Tom Evslin, then recently brought on by AT&T from Microsoft during the mid-nineties proclaimed that if anyone was going to cannibalize AT&Tʼs voice service revenues with IP Telephony (what was then called variously IP Phone, or simply I-Phone), then it would be AT&T itself that was going to do it, and no one else. Theyʼve had ten years to inflict said cannibalization on themselves. Have they succeeded yet? Odlyzko: Yes, AT&T is planning to become a MVNO, but this is largely in order to be able to offer a complete bundle of services to its customers. There is not much money in being a reseller of somebody elseʼs service in general. There is nothing wrong with trying to garner revenues from voice minutes. My argument has been for years that the wireless industry has been negligent in ignoring the opportunities in more wireless voice and higher quality voice. Once we get to real ubiquitous and inexpensive wireless broadband state, that opportunity will be gone (as it is pretty much gone in the wireline area), but right now it can still be exploited, but these guys are still mesmerized by content. AT&T has done very little cannibalization through IP Telephony. that transport/network costs are already largely paid for in the case of DSL/coax. More importantly, especially with cable, it is very inexpensive to offer higher speeds. In Canada standard cable offering is now 5 Mpbs is $C 44.95 per month, broadband “classic” at 1.5 Mbps is $24.95 and broadband light is $12$19. Cable and DSL Internet Access Cost Structure - an example $ Per customer per month for model network build: Total Costs xDSL $47 Cable $40 Transport/network xDSL 2 Cable 4 ISP/Hosting xDSL 5 Cable 5 Customer Acquisition xDSL 15.5 Cable 14.0 CPE xDSL 4.5 Cable 3 WiMAX Problems again Home Installation xDSL 5 Cable 6 p. 81 Forster: Iʼm a little skeptical about WiMAX as well. I assume the technology is pretty good, but anywhere that has two decent broadband providers will be a tough area for a third entrant. It might do OK in un-served areas, although there it will have to compete with lower priced Wi-Fi gear with meshing capabilities. Service/Billing xDSL 11 Cable 7 Maintenance xDSL 4 Cable 1 In the long run fiber and wireless combinations are a winning combination, but in the meantime DSL & HFC Cable/ DOCSIS should pretty well saturate the broadband market in much of the US and other similar markets. St. Arnaud: I agree with Jim that I donʼt see a business case for fixed point-topoint wireless in competing with DSL or cable. A good web site http://www.cybertelecom.org/data/broadband.htm shows the costs of DSL and cable deployment. It is important to note that all these same costs would apply to a wireless provider, but I suspect that CPE equipment and maintenance would be higher for fixed wireless. The only difference is 164 Sege: WiMAX will probably be good for rural areas and small business where cable/fiber does not reach today. Note the success that Towerstream seems to be having in the Northeast with preWiMAX into businesses today. p. 83 Odlyzko: Apropos the last paragraph, I agree. I have been saying for a while that the number one imperative for service providers is to teach their customers how to increase their traffic. And some are able to do it. I was in Korea a few weeks ago, and their traffic (as well as that in Japan) appears to be growing close to 100% per year, whereas here in the U.S. we appear to be down around 60% a year. VoIP and Security in Government and the Enterprise pp. 84-85 COOK Report on January 27: The COOK Report on Internet May June 2005 http://www.computerworld.com/newsletter/0,4902,99258,00.html?nlid=PM JANUARY 26, 2005 (COMPUTERWORLD) - A new report from the National Institute of Standards and Technology urges federal agencies and other organizations to take care in switching to voice-over-IP technology because of security concerns. The 99-page NIST report, “Security Considerations for Voice over IP Systems,” includes nine recommendations for IT managers to help them implement VoIP in a secure manner. “Lower cost and greater flexibility are among the promises of VoIP for the enterprise, but VoIP should not be installed without careful consideration of the security problems introduced,” the report says. “Administrators may mistakenly assume that since digitized voice travels in packets, they can simply plug VoIP components into their already-secure networks and remain secure. However, the process is not that simple,” the report says. Symposium Feb 6-24 Skype is Like Apple II in the Enterpise p. 88 Reed: Skype is kind of like the Apple ][ in the enterprise. People are using it, but IT doesnʼt like it one bit. Iʼm sure the Skype people want to see it viewed positively in the enterprise. Iʼm sure the Skype competitors want to spread unreasonable FUD, and also justifiable skepticism, in order to keep Skype out of the customers they think they rightfully “own”. The best way of course would be to make their service work as well as Skype out of the box, and not tie it to enterprise sales [As I left Lotus in 1992 to join Interval Research, one of my parting warnings was that Lotus Notes *must* be purchasable and operable as a solution that wasnʼt just an “intra-company” solution, or they would lose big to the Internet. I donʼt think they listened carefully - they continued to create private networks for customers, and created boundaries for inter-corporate Notes connectivity that were operationally too hard to surmount. This failure was one of the sources of the “Reedʼs Law” idea, since it stunted Notesʼ potential precisely due to the lack of paying attention to how group-forming creates value]. [snip] Coluccio: “What to do?” is probably not a question that any individual can answer for the whole, but rather, “what will folks do?”, is more apt a question to ask here, where matters or security, if not overall architecture, are concerned. And through all of the complexity that is usually ascribed to the newly engineered, deliberate forms of kludges that result when additional boxes are inserted in the diagram, in order to assuage the problems brought on by the other additional boxes, may there not be an added level of security from viridae and worms sent by the bad guys that now have to negotiate those additional boxes, than a purer form of end to end model would present? Although this sounds silly, the question is legitimate, in my opinion. Iʼm just asking some questions here that happen to be surfacing in my mind as I read through the list comments, without any prior suppositions of what the answers to those questions might be. From the standpoint of doing any type of business in a commercial context, voice and video conferencing applications may be viewed as being as far removed from the architectural ideals of the Internet as a computer numeric control application is on a factory floor. Itʼs merely a business tool that has to work right when called upon every time. And they are nothing more or less than that, from the standpoint of most who use them. p. 89 Davis: Well said, Frank, at least from my perspective. I am an engineer and an architect. I am also the one my company trots out to government and enterprise customers to give the hows and wherefores . . . customers who are highly concerned about cost, availability, insist on the telco-like SLAs to which they are accustomed (four ninies). Frequently 165 there are competing customer groups: the older, telco-accustomed, frequently proxies for their customer favorites and personal relationships built up over decades, and on the other side, younger “whiz-kid wana-bees” who tout their protocol religion and vendor of choice. These are real issues up at the political end and tribal end. I am the one usually brought in at stalemate stage. These larger organizations, unlike individuals who are free to care or not, who are constrained by statute, case law and GC interpretation as to liability and contracts, and policy to provide at least disciplined due diligence on security. At the Operations level, the security issues are not only of privacy and confidentiality of digital information asset stores, but the requirement to keep the network up and operational, especially as the post-BUST cost cutting has driven staff to what I think is excessive thin-ness. The interoperability issues that Frank raises are essential in designing and operating. As a simple case, the Geoffrey Moore curves on market adoption of new technology hold true. The economic case is rarely won by the swift. The Middle and Late Adopters are the huge bulge in the frequency chart that is definitely negatively skewed. To meet the “four nines” SLA, absent any, using Gordonʼs term, “disruptive” and rapidly adopted by a super-majority of the economically significant population, signaling or other in-band or out-of-band IP protocol, such as Yakov Rektor and some of us techno-agnostics at Cisco attempted to promulgate, the SS&/AIN network is essential. Government-to-Business, Government-to-Citizen, Business-to-Small Business, and Business-to-Citizen will require it. To some objectors, citing that IP was designed to be connectionless, I respond that TCP was designed to be “connection-oriented”, just as ATM is connectionless but “connection-oriented.” To those who insist, with what I always hear as protocol religion, on EoIP (nothing less than Everything over IP), I respond, COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA and Yakov responded, that IP was never designed to do anything with voice or video, whereas ATM was, and . . . policy. We had to do the demo at a site outside the corporate campus because of their overwrought policy. if the engineering and architecture of communications networks is about “the communication” part - User Adaptation Layer, letʼs work on that problem, using whatever tools we have or can design in the process. p. 90 Reed: Frank - I wish that the internals of enterprise networks were just part of the Internet. Melissa - who worships no protocol godlet. Reed: Donʼt bet on the Enterprise markets for any of this, Gordon. Enterprise use of new technology is now dead. The Chief Security Officer outranks the CIO, and each one says: “Be afraid of novelty, be very afraid.” Just imagine your worst nightmares and project them on new technology because itʼs weird. This just reminds me of 1986 at Lotus when the Chief Security guy (our corporate counsel) said that we should not allow our employees to interconnect our electronic mail with the Internet, because that would be the death of Lotus due to the raging security risk. As a vice president in charge of developing technologies and products around (among other things) electronic mail products and stuff like Lotus Notes, I and others argued that we couldnʼt afford not to interconnect. This was not an easy fight. There was “no business benefit” to interconnecting to the Internet in those days - we were told we had to limit interconnect to a small number of essential R&D personnel who could demonstrate a “need to email”. When we engineers said: “No, Jim Manzi should be using Internet mail.” People thought we were joking. Jim, fortunately, understood the point, and chose to make an example of himself. It also reminds me of the time Starwave and some Interval Research people demonstrated the first commercially successful Web service to Microsoftʼs executives (ESPN Sportszone) in late 1993. Microsoft could NOT connect to the Internet, again because of corporate However, as I mentioned, in many companies, the CSO has ordered that use of contact managers (like Plaxo), VoIP (like Skype or Vonage), access points, Wi-Fi, ... are firing offenses. And to back it up, various gear providers are providing traps for such “evil technologies” that can be deployed on corporate campuses. So yeah, those of us who live largely outside the “enterprise” as consultants, pundits, small-business owners, etc. donʼt see this phenomenon. But I can assure you, you may be able to technically route Skype out of your PC on the corporate network today, but the trade organizations of corporate legal departments are telling all their members to make policies against it, to require technology to block it, and asking senior executives to direct against it. After all, you wouldnʼt want your employees actually to be empowered would you? Thatʼs a scary thought! This fear of novelty is one of the many reasons why genuinely disruptive new communications technologies come last to the Fortune 500. The enterprise market eventually gets there, but itʼs rarely in the vanguard of anything. Coluccio: David - Just for clarity, Iʼm Skyping from home, not my clientsʼ offices. ;-) On that note, however, I recently spoke with one my associates who IS situated behind a client desk (and who also VPNs from home two days out of the week), and I asked him if the clientʼs security hawks were still adamant about prohibiting IMʼing over the corporate LAN, as they had been a while back. I was advised that they are now using an in166 house IM package, which, while capable of talking to other vendors/SPsʼ versions of IM, still cannot do so for the types of reasons youʼve cited and otherwise implied. Also, Skype is not to be seen anywhere on the horizon in that particular account, while there are staff that I have situated elsewhere who say that itʼs being used with mixed resistance, still, and Iʼd imagine that the smaller the number that exists following the name FORTUNE 500, the more resistance one will find. p. 91 All of my big customers are doing test beds or actual deployments of Ciscoʼs VoIP or Nortelʼs VoIP. Vonage is, among my customers, in no way singled out as a target. I wish you would explain the point of why you wish, as you state below, that “the internals of enterprise networks were just part of the Internet.” Why should they be? The internals of the enterprise networks are privately owned and operated, under the constraints of the marketplace of P&L, for the business requirements of that business/enterprise. Those enterprises are governed by statutes (Sarbanes-Oxley, HIPAA, etc. Government networks must meet the requirements of Clinger-Cohen, e-Commerce, e-Auth, and coming is HSPD-12.) All businesses are bound to protect privacy and integrity of both financial transactions and employee personal information (personnel, discipline, evaluations, labor grievances of some kinds, EEOC violations). Businesses and Government networks institute policies for very secure authentication and communications with and about employees with some risk, likelihood of being targeted by hostile agents (kidnapping, etc). Businesses and Government networks must maintain privacy, integrity, and transaction logging of sensitive management, trade secrets, intellectual property, strategy information. I would like to hear your dissertation on the compelling reasons for opening all that up to any old curious person, or expose that non-public but legally obtainable information for public discussion. The COOK Report on Internet May June 2005 I will tell you that I am, with others, engaged on ferreting out an application you did not mention, specifically “GoToMyPC”: a remote control program that defeats and renders useless enterprise firewalls, IDS/IPS, and other packet inspection devices. GoToMyPC, once installed on an internal enterprise machine, maintains an open outbound connection HTTP (443) to a remote proprietary GoToMyPC server and exchanges shared private keys with AES128. The remote agent contacts that broker/server from some outside point with unknown and unknowable authentication standards (weak passwords, e.g.) This places anyone with the legitimate credentials (possessed by a user for use on the internal network or possessed by a poseur who has stolen the credentials via RAT/keystroke logger, or zombie via RAT) in complete control of the internal machine, its access to other machines and network stores. Internal security policies have no effect on restricting the export of internal information as, to the network devices, the request is received from a device internal and local (the invisible remote user not being visible). The encrypted connection between the internal machine and the remote machine through the GoToMyPC broker is opaque to any known sensor or firewall device. Every customer I have wants “empowered workers” without themselves becoming disemboweled in the process. VoIP presents special problems in the corporate/enterprise/government environment. Business cases have to be made, risk assessments done, mitigations proposed-tested-deployed that meet cost-benefit decisions. Information may be free, or practically free on the Internet, but information is not free in the corporate/enterprise/government environments. There, information costs to produce, and that cost has to be justified with some judgment of ROI. The NIST report - Yes,Gordon, the 100 page one- lays out the risks. No one says the risks are proscriptive. One can, and CSOʼs do have to accept risk. p. 94 Coluccio: Narrowing the field of ʻcustomersʼ down to just three, i.e., Consumers, telcos, and enterprises is in my opinion an over simplification. What happened to the large base of government-based networking taking place, not to mention an even larger base of enterprise customers, municipalities, greenfield operators, condominium builds, etc? To further break it down, some enterprise customers are asset-based network operators unto themselves (some in a very large way), meaning that they own major portions of their own infrastructures. I submit that in each of the categories of ʻcustomersʼ listed above you will find both SIP and Skype infiltrating into traffic flows, whether they are sanctioned or not. Enterprise IT departments fill proxy roles similar to that which David refers to when he speaks about the vendor-telco, or vendor-consumer relationship. In the case of the enterprise, IT is both service provider and the vendor that counts, and has the last say in what is authorized and what is not. So, IT folks, too, have to listen to their ʻcustomers.ʼ And while there are many applications that make sense, like IM, those apps donʼt always become endorsed or supported by IT in a timely manner. If they are compelling, they eventually find their way onto usersʼ nodes, even it if means putting the enterprise logo on them first. And this is exactly what Iʼve seen happen in my client organizations. And in a way that weʼve seen this take place with IM already, where the Merrill Lynches (I believe), the JPMs and many others have implemented it, weʼll very likely see the same thing occurring with a -like application before long, if not licensed directly by and others, themselves. You asked earlier if Iʼd offer some examples of enterprise applications that might be conducive to , perhaps in banking or brokerage. How about junkyards? Junkyards and financial trading floors share more than one legacy together, you know :) 167 One is a form of public address system application that is tied to a conference circuit arrangement known as a “hoot and holler” network, or, variously as, “shout down” circuits, or even “order wires” in some folks books. With some tweaking to ʻs conferencing features I can see where similar functionality could be achieved, although I donʼt know just how far tampering with it would constitute creating another application altogether. See one vendorʼs approach to hoot and holler over IP, below: http://www.bsslimited.com/Business/ KCS_IP_Hoot.html Cisco has done considerable work (as have others) in adapting this application to IP, as well: http://makeashorterlink.com/ ?B6EF5267A The Importance of Directory Services p. 96 Stastny: Regarding Gordonʼs three categories [of broadband], - I would put municipal networks under enterprises, as they have essentially the same requirements The 3rd point can be split in two: a. Enterprises connected to telcos b. Enterprises wanting to peer directly Case a is supported within the sipconnect group that is found at http://www. sipconnect.info/mc/page.do with its interconnect draft in turn found at http://web.memberclicks.com/mcdatafiles/site/sip/SIPconnect_Version_1_ Draft_2-2-2005.pdf Case b may the most important use case for ENUM Fixed Telcos and mobile operators are currently coming back to the market of VoIP very strongly, both in the US and in the rest of the worlds (see 3GPP/TISPAN/ATIS using IMS, but finally the whole issue boils down to the question: will the consumer need telcos anymore COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA or not? Note that the SIP server manufacturers play a crucial role here, they may even be betting on the wrong horse. The horses they should be attending to is the device and end-user application providers. Device manufacturers basically do not care who is buying their equipment, just look at the mobile phone industry Of course they like to sell in bulk via service providers, but they also sell to the end-users. And any general-purpose mobile phone or GSM-enabled PDA is another step to the death of the service providers. The big US carriers, the big mobile operator groups and also the cable operators are currently competing like mad within and between the groups and they seem completely forget Clay Shirkysʼ Zapmail example: http://shirky.com/writings/zapmail.html You can compete with everybody except with your customer So this battle will be decided from the customer, and the first one really satisfying the customerʼs needs will succeed Skype is well ahead here Coluccio: Hi Richard. I had to read your message twice before I realized that you would have been much better served in my opinion if you had prefaced it with the following qualifier: “From a directory services perspective, ... “ Stastny: Hi Frank. Sorry, I am currently so immersed in addressing, numbering and naming issues that I assume automatically everybody else is also ;-) Coluccio: Thereʼs hardly any cause to sound an apologetic tone. The single most rewarding aspect of participating on this list, if not challenging at times, as well, has been the kaleidoscopic effect itʼs had for me and Iʼm sure others, seeing the world through so many other professionalsʼ eyes on a sustained basis. I certainly appreciate your viewpoint and those of others, even when connecting the dots, using both real and virtual links, sometimes requires cracking a dust collector, or two ;) P. 97 Henshall: Martin Geddes wrote a piece on [Goroshevskyʼs] Popular Telephony and Peerio that may be interesting to you. Generally the jury is still out. Are they vapor or not. Iʼm yet to see anything ever. Still they get press, I donʼt know why. The ideas are interesting. See http://www.telepocalypse.net/ archives/000602.html pp. 102 Sebastian Hassinger: I believe that Skype is just another sign of the inevitable restructuring of the communications sector from a vertical to a horizontal orientation. Whereas the landscape has been historically been dominated by vertically integrated firms who own and operate networks, selling bundled connectivity and ʻvalue-added servicesʼ to its subscribers, we are moving towards horizontally layered network providers (various flavors of broadband, wireless or wired) with the ʻvalue-added servicesʼ (applications) provided by unaligned firms that ʻfloatʼ on top of the networks. This is a natural evolution for maturing networks, and in fact one could argue the communication sectorʼs transformation is way overdue, retarded by the incumbentsʼ reluctance to change and the regulatory deck that has been stacked so decidedly in their favor. One reason this decoupling is inevitable is the economic reality that the optimal business models for selling network connectivity and network applications are at odds with one another. An operator wants to drive as much high value traffic as possible across its network and an application provider wants to reach the broadest audience possible with its service. Therefore, a vertically integrated operator/provider is always internally conflicted - the network business wants its value-add services to be bound to the network it owns, in order to increase the traffic and the value of that traffic. The value-add service (application) business, however, doesnʼt want to be bound to the network because that limits its market reach. Imagine a -like application owned by 168 a DSL or cable provider that favored its own network - allowing free calls only to subscribers on the same network, for example, and charging for calls that ʻleaveʼ the network. Imagine this network-bound -like app competing against the actual Skype with its polymorphously perverse disregard for the underlying network of its users. No contest. In fact this is the often the case with the features of the mobile networks today - free calls to other subscribers on the same network, or Push-to-talk only interoperable with phones on the same network, or even the spotty operation of SMS across networks. There is a robust business in operating ʻcommodityʼ networks that are agnostic to the applications that push bits across them, but it is a far leaner model than that practiced by the vertically integrated OpEx heavy firms of today. Some will make the transition gracefully, some wonʼt. p. 103 COOK Report: Are we indeed headed to a world where the only thing monetizable will be bandwidth? Forster; No. While I agree that Skype, and VoIP in general, are adding to the pricing pressure on voice-type services, I donʼt think bandwidth will be the only thing that will be monetized. Whenever an amorphous and ill-defined service that is done somewhat differently for many sets of customers at great expense can be crystallized into one thing that can do 90% of what everyone wants very conveniently and at low cost there is a great opportunity for profit. What seems to be different is that for telecom the service used to be defined in terms of what came out of the wire (telco demarcs), or how the device behaved (cell phones, IM devices, etc.), but now the only constant is the IP packet and IP address; all the rest is in constant motion and the result is a lot more like software. Adding to the confusion, sometimes the software is momentarily solidified into devices (Skype in a phone). p. 104-105 Matson: Gordon - Many of us think that the major private enterpris- The COOK Report on Internet May June 2005 es simply havenʼt begun to understand the implications of flat, universal global peer-to-peer communication has for their business models and very structural survival. I am not sure there is much to be gained by “blue-skyping” the future in this manner, but I have for years held to the words of the great Walter B. Wriston: “”The philosophy of the divine right of kings died hundreds of years ago, but not, it seems, the divine right of inherited markets. Some people still believe thereʼs a divine dispensation that their markets are theirs - and no one elseʼs now and for evermore. It is an old dream that dies hard, yet no businessman in a free society can control a market when the customers decide to go somewhere else. All the kingʼs horses and all the kingʼs men are helpless in the face of a better product. Our commercial history is filled with examples of companies that failed to change with a changing world and became tombstones in the corporate graveyard.” Maybe Wristonʼs wisdom applies to a much higher level of the established order than just individual “companies” and “markets” in times of seismic technological change such as we are currently in. Davis: Malcolm, I totally agree with both of your statements above, but not the implication lying transparently in your syntactical construction, some indicative mood form of a French “nʼest pas” implying that the audience should bow in affirmation. 1. I havenʼt seen such business models. 2. Enterprises (private, public, government) have business models globally and very distinct component models, each with differing levels of need for the CIA triad (Confidentiality (sensitivity), Integrity, Availability). If your statement and your quote are simply techno-religious a priori value axiologies, of the now trite cliche of “Internet vs Telecom”, or “Power to the People and tear down the walls of the oligarchs”, or a moral imperative for the kind of transparency that makes long haul optical glass seem opaque: then we simply have nothing more to say to each other. I donʼt share your axiologies, therefore discussion is impossible. The eco-spaces of end-users and their communications needs are fundamentally different from those of complex organizations, though for some businesses overlapping at the edges. Enterprises, whether privately held, publically held, or governmental have external constraints to protect and maintain confidentiality and integrity of certain information stores and transaction records. Consumers are not so constrained. Enterprises are playing in a competitive environment. Intellectual property, business strategy, and much of corporate governance must be maintained with restricted availability to “need-to-know.” The cost of loss of availability, as to propagating worms, is huge in any business, where, for the end user, loss may be inconvenient, but of minor financial impact. Per the following quote from the Skype web page: “We prefer to think of ourselves as a big group hug, even a present. Yes… thatʼs it… weʼre a present… but without the ribbon.”: sufficiently how Skype, and other such “firewall-friendly” applications work (also reference “GoToMyPC”). As David Reed has said often here, the insider threat far, far outweighs the outsider threat. Skype is not penetrating the firewall from the outside in. Skype penetrates the firewall from the inside out, on the same TCP Port 80 or 443 connections that are open by design for web browsing. Skype goes out just as your Firefox or (hopefully not) Internet Explorer goes out to LightReading or CNN or Google. What is different about this breed of applications (and GoToMyPC) is that they maintain persistent strongly encrypted connections to external untrusted or untrustable servers. That may not matter to an individual end-user or a “hot-spot”. It does matter to any organization that is liable for at least certain categories of its data stores (by statute or to stockholders), privacy of personnel and medical records, employee behavior (e.g., harassment). It may seem “way over the top” for a corporate or government agency to have to acquire, secure, and maintain TBʼs of email stores, but the cost is far worse for not being able to answer the subpoena. My customers and I retort: “Beware of geeks bearing gifts.” Reed on March 8: Melissa - your point that connections to from employeesʼ machines to outside resources pose a risk is dead right. Skype, Firewalls and Security But the risk is not exacerbated by those connections being persistent or encrypted. Editorʼs note – this exchange between Melissa Davis and David Reed came in 2 parts over a space of two weeks. I excverpt it in full because I judge it to be extremely informative. The real problem is that internally companies are “wide open”, so any random person passing through (especially employees and their visitors) are exposed to information they shouldnʼt be allowed to see. pp. 105 – 07 Coluccio: Also, what of my earlier implied observation above concerning Skypeʼsʻ ability to penetrate WiFi hot spotsʼ firewalls, even when those hot spots are not part of such a promotion or ongoing service? Davis: Frank, perhaps I havenʼt explained 169 Making rules at the membrane surrounding the company is like creating the Maginot Line - you bet your entire security on the assumption that the path the attacker will take is where you think it is. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Essentially this is because keeping the internal records secure *internally* is viewed as “too expensive”. This unwillingness to do the job right shows up in lots of ways. Why was that “backup tape” of Bank of America payroll records that made the front pages last week not encrypted when it was shipped on a truck and insured for a few hundred dollars? Surely the BofA risk managers know that exposing such personal information is important to stop. Similarly, you wouldnʼt have to worry about people making phone calls through the firewall (even if the phone calls could be hijacked by VoIP hackers to carry other info), if the fundamental security principle called “the principle of least privilege” were applied consistently throughout organizations. Protecting an organization against losses should not be a matter of random stabs in the dark, worrying about symptoms and highly-hyped threats, but instead should involve keeping the valuable and highly private information inside an organization in a secure manner in the first place. In other words, implement a security infrastructure *inside* the organization that limits access effectively. Then Skype or VoIP or whatever cannot be a threat in the first place. Davis: David, You are dead right about everything you have said, with a single exception that I, at this moment, disagree, but am willing to be further informed by your argument if you will advance it another few steps. Point I wish further clarification on immediately below: David Reed wrote: But the risk is not exacerbated by those connections being persistent or encrypted. But first Per agreements: 1. Simple “Maginot Lines” of perimeter security are of little usefulness, for enterprises with the beaucoup bucks they spend on SPF/FW devices. 2. Inside the “RED ZONE” (Public Internet) perimeter are the email and web packet content inspection and filtering engines, the publicly facing Directory Servers (Border Directories), SIP Proxies (if exist), Web Proxies, and IDS/IPS (Intrusion Detection Systems/Intrusion Prevention Systems), TACACS+ and RADIUS RAS (Remote Access) services. The Citrix Nfuse and ICA servers would sit here, if exist. For medium companies, this is a single DMZ. 3. For larger enterprises, the description in #2 above, exists in an Orange Zone DMZ, handling traffic primarily facing the Red Zone, or the Public Internet. The main ingress and egress MTAʼs would sit here. 4. Behind the Orange Zone is a Yellow Zone DMZ, primarily handling traffic and transactions within and between the various Green Zones (considered Trusted Zones). Intranet Web Servers serving as secure portals for internal transactions (e.g., finance, HR, labor, executive strategy) would be here, Yellow instead of Green if the Enterprise wishes remote access. Or they may be in Yellow because that is where the interdata center links terminate. 5. In side the internal collision domains are internal SPF/FWʼs and stealth mode IDS/IPS sensors. The main function here is to monitor the presence and epidemiology of worms, port scanners, unusual activity. The legal “procedural defense” rationale is to establish that user behavior is subject to monitoring. Now, you and I know that packet sniffing is a notoriously bad way to do something like that, given the sheer amount of traffic, but it is arguable and is argued. 6. The IDS/IPS systems are also there, along with the NMS (Network Management Systems) to map traffic in effort to discover “rogue gateways” (illegal or unauthorized gateways to the Public Internet). 7. Internal NMS monitor switch and especially router interfaces and buffers, 170 again to detect the kinds of congestion or malformed packets that would indicate a potential packet pandemic condition. 8. Internal workstations and mobile laptops have virus scanners on them. With internal workstations, the scan engines and signature updates are pushed out with software transparently to the user. 9. Laptops are a huge problem and are getting more and more attention as a political battle focuses on locking them down vs “creativity and innovation.” Increasingly, mechanisms are being planned and experiments being done with requiring scanning before the login is consummated. That is, with a login request, the user request is placed in a queue while another process patches and scans “out-of-compliance” machines. 10. All of this, as you know, should follow a thorough assessment of what infrastructure and data resources an enterprise has, triaged by risk (impact of loss or compromise). It is here, David, that your major point of “least privilege” applies. Is it done? In my experience, yes when it comes to infrastructure (routers, switches, server sysadmin). My experience with application access, and most particularly data store access, is that it is much easier for a business unit steward to grant privileges than to revoke them. If there is a formal employment adverse action, the revocation is not so difficult. Absent that, and particularly in “white collar union” shops and civil service, one could be challenged to meet almost criminal proof standards to sustain such a revocation. Further, particularly North American enterprises have largely followed by a long way that of major European enterprises in adopting X.500-like LDAP Directory Services, which allow authentication to be done globally with authorization done locally (with a few mouse clicks). Consistency checks and date checks can be run to produce alerts on possibly stale permission attributes. The COOK Report on Internet May June 2005 David Reed wrote: But the risk is not exacerbated by those connections being persistent or encrypted. Davis: David, here is where I may be blinded by my own training and the cybersecurity culture in which I immersed myself after the Bust. 1. the persistent, always on, connection seems to me vulnerable to any number of hijacking, session key theft, and intrusion techniques through known exploitable applications, e.g., Windows Media Player. 2. the persistent connection is simply out there hanging for any black hat to use as a route of propagation for any kind of malware they have, again, Windows Media Player is an excellent example of a propagation route 3. if I allow a non-encrypted connection, even locally to a proxy, I have the snooping ability to investigate suspicious activity (an insider exporting protected information or a “captured” insider machine with RATs, Zombies, Trojans, Data-Miners and other digital critters. 4. perhaps not in general with , but definitely not with encrypted sessions, can I enforce what I am compelled to enforce, specifically calls which are sexually harassing, conspiratorial, etc. Please tell me what I am missing. Reed: Melissa - Disclosure doesnʼt require persistent connection. A novelʼs worth of data (plenty if you are trying to capture a corporate strategy) is only half a megabyte or so. Given the number of bits flying in and out the doors of any company, that takes an instant. It fits in a tiny corner of an otherwise innocuous document, perhaps by misspelling occasional words in emails over time. It takes a few *milliseconds*, not hours. Disclosure doesnʼt require obvious encryption. At the network level, you or anyone else cannot understand what 99% of the bits mean, or understand what bits are reasonable and what ones are not. It may be that you think you are mandated to listen to everyoneʼs phone calls because “sexual harassment law” requires you to do so. I understand that thatʼs the interpretation of many lawyers. Independent of my judgment of whether thatʼs good public policy, itʼs also technically stupid. You cannot assign enough people to the job to notice and block all attempts at sexual harassment merely by listening to phone calls. Even if you hire a whole staff of outsourced Indian English-speaking listeners at low dollars per hour! The same argument applies to the idea that you can monitor every possible word flying through the network and prevent theft of information. The answer to blocking theft of information is to control the flow of information to the small number of people who need to know in the first place, not buy thousands of dollars of equipment because you want to buy Microsoftʼs garbage systems and patch them around the edges because they do not allow you to build appropriate security policies at the right grain in the first place. Davis: We have no disagreement here. Your words below capture the problem eloquently. Reed: The answer to blocking theft of information is to control the flow of information to the small number of people who need to know in the first place, not buy thousands of dollars of equipment because you want to buy Microsoftʼs garbage systems and patch them around the edges because they do not allow you to build appropriate security policies at the right grain in the first place. Davis: As for the monitoring, again we have no disagreement. As a practical matter, no team of people is large enough to monitor the volume of transmissions and content filters are easily fooled with syntax (e.g., double negatives) mis-spellings, and shared secret word substitution. 171 In the case of suspicious behavior, with internal crypto, such as PKI, the stream could be encrypted and the transaction time recovered under legal or other policy permission. But for now, the main points are those of the lawyers and the “procedural defense”, as well as the fact that these enterprises continue to pour big dollars into defective insecure operating systems. But Skype May Show up in Some Enterprises Sooner Rather Than Later pp. 108-09 On February 20 Coluccio: Over the past two weeks, on an unsolicited basis Iʼve become aware of two instances, merely by speaking with acquaintances, where Skype is actively being piloted by a very large business enterprise. In one case, I was told of a publicly traded semiconductor manufacturer that has discovered some flowrelated issues that have arisen inside its firewalls, which is looking at Skype and should have a fix on, soon. In another, a financial institution on the East Coast is readying a launch of an application similar to Skpye (which was hinted might even be Skype dressed up for the occasion as something else), but no word as to when that might be. Shockey: Iʼve heard this as well. Skype is spreading like wild fire in European circles. James Enck the noted telecom analyst with Diawa Securities has noted this extensively in his blog http://eurotelcoblog.blogspot.com/ Coluccio: And clearly, Skpye has had a very active front office, despite images one might conjure by statements in the press about its limited size and presence, as witnessed by the number of very serious announcements that have come out over the past week, alone: a broadsweeping announcement with Motorola; Amazon and Wal-Mart will be hawking it i.c.w. a PC bundling arrangement with a Unix OS and Firefox; and discussions surrounding a host of mobile devices COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA that would become Skpye ready, soon, if not already. governmentʼs concerns over security and reliability. IM was once regarded with the same level of skepticism and mistrust in large enterprises as Skype receives today, but over time it has gained acceptance, even at the cost of losing its identity during screen pops, yielding to the monikers of the firms who employ it. And in those cases where IM is used, in some cases by the nationʼs largest banks and brokerage firms, they are indeed designed to operate as walled garden applications, barred from receiving or transmitting anything to or from the outside world. Shockey: See the Microsoft LCS platform. What you described is being delivered. Shockey: Excellent Frank. I recently wrote a magazine piece pointing this out. Skype, Web Services and Mission Criticality Coluccio: Having said that, during the near to intermediate terms, I cannot see either IM or Skpye being used to fill the shoes of mission critical applications or any of the more advanced requirements that are now finding their way into complex Web Services. Shockey: They will once they become standards-based. Web Services have completely taken over the data exchange market because everything around web services is based on real documented standards. Coluccio: Instead, they will continue to be used as a form of expedient to save time and steps to get the job done, by taking the place of email and phone calls. And this will be so NOT because they cannot be made to perform to fulfill the more advanced functions, but for all of the reasons that Melissa has stated. But I do believe that they will find acceptance when they are modified and/or tuned to the organizationʼs specific criteria as related to both security and their boundaries on reach, and will receive use only in the manner in which they were originally intended, for some time to come, and within designated zones of trust. So much for the blue chipsʼ and the Coluccio: For the other ninety some odd percent of the universe, however, they will continue to adopt Skype in the same ways and for the same reasons theyʼve ever adopted anything else. If it works to a userʼs favor, it will be adopted. If not, then it wonʼt. There are tens of millions of small to medium sized businesses out there whose sole connections to the Internet consist of one or two DSL lines or a T1 line or two. Add some tens of millions of consumers who attach via broadband from home, and now from the street via certain forms of wireless, as well. And then we are hundreds of millions of citizens of all types who overlap with at least one of the categories of presence listed above, which means that you can then mark some of those individuals down twice, sometime three times, as candidates for using IM, or Skpye, or whatever the next pop up act is that comes along. p. 112 Davis: Revolutions impoverish everyone, and usually for a very long time - as Japan, Inc has proven. The shock that has rendered the Japanese stagflation impenetrable to monetary and industrial policy is the shock to the culture of the collapse of the real estate and banking enterprises, which were backed by a government/MITI that had so long promulgated invulnerability and wisdom. You are seeing now in the telecom field what has been predicted. You are seeing what your anarchistic leanings fear the most . . . consolidation, or re-consolidation, “pay for play.” IP won as the transport of necessity now. No one cares, especially the ILECs about the circuit switched legacy dinosaur and all are moving to max revenue generating businesses and services away 172 from “common carrier.” FTTH is dead except as a symbol - making almost no economic sense with a single exception - the ILECs see their competition for the SOHO and residential customer as the cable coʼs, and each other as competitors for the large enterprise/government space. Services, the ability to deliver a branded product with a stable provider at a reasonable price, is the game for SOHO and the bulk of the non-early adopter consumer. You forget how techno illiterate the consumer is. In the enterprise/government, bandwidth is simply assumed to the edge. The differentiators are about service delivery and SLAʼs, as well as MTSP (MultiTier Security Planning) deployments that cut the internal costs to the enterprise and fovernment customer. I donʼt need a solution, Gordon. That isnʼt my game, predicting the broadsweeps of some future. I am an engineer and an evolutionist. Evolution doesnʼt reward more than very few of tens of millions of mutations, and those are small ones, mostly neutral in terms of survivability and incremental improvements I donʼt know what optimization even means. I know that what I have is a niche is grey matter, experience, knowledge, and the self-discipline to get more and make that activity fun. So I only think about getting through the next few steps and how to get a decent check for that. Being a player in the “race to zero” is not in my interest, personally or professionally (by association). The Google Telephone Network and a World of Abundance in Communication p. 113 COOK Report: And back to the Google dark fiber discussion (pages 71-76 above). Have you all looked at Googleʼs new local feature and ʻs voice mail? Look at the maps available now on Google. It is much better than map quest. You can query those maps for all hotels within five miles of any address. Why not gas stations, restaurants, in short why The COOK Report on Internet May June 2005 not any kind of specific retail business? And why not the next feature being a Skype or Sype like calling number for that retail business? There is a potential here for removing vast amounts of phone traffic from the PSTN to the Google telephone network. I am just try to see some of the possibilities of where this could go. But the cautionary pushback is a welcome signal to choose my words with care. You have mentioned a race to zero that I find frightening indeed - but it seems that something - like it or not is happening - and the question becomes what are the activities in the society that add value and people will pay for? Or if not this what does the question become? Davis: Gordon, here is what I am looking for: simply a business model that would provide sustainability, scalability, direction, control. In all of anthropological research, there are only a few sustainable models that work. There is the techno primitive fairly flat, family and extended kinship of the subsistence hunter-gatherer, the warlord, the feudal cum nationstate model. There are equally only a few economic models that have shown workability, mass socialism of a Voltairistic model and mass Libertarian models falling to state or economic oligarchy (and apparently in the Western World, so do republics, parliamentary and the US model). Open Source, e.g., Linux, is not the freevolunteer cyber peer-group of yore. Open Source programmers, and “the maintainers” of significance in the community now are largely employed by Intel, AMD, HP, IBM. Linus and McVoy are corporate animals of this consortium. So, my question has to do with the economic model. Given that significant economies are all monetized, what is the financial incentive? QoS OpEx Economics - Flows or UCLP? pp. 122-23 Bill St. Arnaud: Technically this [what the BCR piece above says about QoS and OpEx] may be true, but rarely do you see a systemic comparison of QoS and OPEX versus an alternative of increased CAPEX to significantly increase the bandwidth. The OPEX costs of managing QoS are horrendous and, because they are so labor intensive are only going to increase. CAPEX costs per bit of bandwidth continue to drop dramatically. Many carriers feel they will be a charge a premium for QoS that will more than offset the OPEX costs. But, it is hard to imagine how carriers will recover a premium for a service like voice that has never seen as a premium service by the public. Can you imagine carriers trying to sell a service that ensures greater probability of having better sound quality or a higher likelihood of not being dropped? Odlyzko: Very well put. QoS introduces new levels of complexity into an already complex system. Roberts (to Bill St. Arnaud): Iʼm not sure what I said that triggered your comments about QoS Capex and Opex. However, I am happy to comment on the issue. However, to clarify what my comment was about, it was that Wi-Fi or 802.11 has serious problems with delay variation in practical use. I only had same small private experiments but the paper referenced by Henning Schulzrinne confirms my concern, only a few (10-30) VoIP calls are possible over 802.11b and with normal computer FTP, very few will work acceptably. 802.11g would reduce the delay, but has a much shorter range and thus coverage. This should be of great concern for cities that are installing 802.11 across the city if they think it will support WiFi phones. The same holds true for Motorola and Skype if they hope to use cheap phones using 802.11. Now, back to your economic point, Bill. I now believe that routers built to route flows rather than routing every packet, eliminate a vast amount of complexity, reflected both in CAPEX and OPEX. When flows are routed, it is easy to manage the QoS of the flow. The more the flow is like a fixed rate TDM flow the easier is it is to manage and support low loss and low delay variation. The more a flow is like TCP with big bursts and un173 controlled growth, the harder it is to support and manage. Perhaps the higher cost of managing TCP over VoIP can be offset by the use of cheaper excess bandwidth, but overall, as you know, fiber bandwidth is the lesser of the costs today. So I would contend, that bit for bit, a constant TDM stream like VoIP should be cheaper than WWW or FTP bits. I cannot predict the charging model that will be used, I am only looking at the CAPEX and OPEX cost. Carriers will charge what the market will bear, and if voice sells at a premium, they may charge a premium. But as flow routers become more widely used and the competitive market settles down, the price per bit may be the same for any QoS. That is because, managing QoS only requires discarding and scheduling correctly, two much easier things than routing and DOS on every packet. Done correctly, there is no increased OPEX to set up or manage QoS. It is a specified by the sender, either in the DiffServ mark or in other packet information. I realize it will take time for me to prove these points, but do not assume that the structure of routers has to remain as it has for 30 years and that with a new design, that the cost and complexity does not need to appear in the same place as before. I know full well you understand this point. So please accept the possibility that QoS can reduce cost, not increase it. I will prove it to you very soon. (Andrew, the same for your comments). I really do worry however, about QoS at the wireless edge, since the 802.11 protocol is imbedded in the wireless devices and cannot be improved easily. People may assume the same performance as on the wired network, and this is not yet the case. St. Arnaud (to Roberts): It is not the technical costs of OPEX that are my concern - it is the human costs of managing SLAs and billing systems that QoS entails. In the wireless example, there is no question that todayʼs WiFi has limited capacity to support QoS. One solution is to implement QoS so that those customers who have contracted with you to COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA provide a service are guaranteed priority when they make voice calls over your WiFi system. The other approach is to install a lot more WiFi nodes. Assuming everything else being equal, for a given number of Wifi VoIP customers, which would be the cheaper solution? This analogy can be extended to so many other network examples where it is perceived that QoS is the only answer. There are basically two types of QoS systems: internal and external An internal system is one that may be used by a service provider at the edge of a network so that can prioritize traffic within their network - but not offer SLAs to their customers. This is something MSOs want to deploy in order to support triple play on the cable system between the head end and the customer. This makes a lot of sense and is relatively inexpensive An external system is where a customer needs end-to-end QoS and wants a SLA. To my mind this is very costly and expensive because of the SLA management issues regardless of whether you use flows, MPLS with premium service, WRED, TDM SONET or other techniques. However, an alternate approach is to give the OPEX knobs to the customer so that can deploy and manage their own QoS. This gives the customer what they want without incurring a huge OPEX overhead for the carrier. This is what UCLP provides - by using web services we can let the customer control their specific knobs for managing and controlling end-to-end QoS. This will work with flows as well. Technical Aspects of VoIP Traffic Shaping on Wi-fi Network pp. 124 –26 On February 26, David Reed replied to Larry Roberts technical 802.11 question: Larry - just to clarify something you said. The 802.11 standard supports slotted Aloha operation, but it is wrong to say that it *is* slotted Aloha. 802.11 supports both DCF and PCF operation. DCF (distributed control function) is slotted aloha, and suffers the problems you refer to. PCF (point control function) is a centrally controlled polling protocol, with every transmitter getting turns allocated by the so called “access point”. Actual 802.11 networks - the ones you call “cheap”, even the “cheapest” ones - actually use PCF whenever an access point is present. Even the $100 access points use PCF, and tell their clients to use PCF. Schulzrinne: Iʼm sorry, but this is wrong. All access points today use DCF, with PCF rare to non-existent. As far as weʼve been able to determine, Cisco access points are capable of doing PCF, but weʼve never seen it being used. (Also, DCF uses CSMA/CA, which is related to Aloha, but not the same thing. In Aloha, the sender canʼt listen to the channel; in CSMA/CA, it does.) This makes sense since data transfer performance for PCF is dismal since a large fraction of the bandwidth is wasted on polling. Reed: Quoting from the 802.11 Handbook (IEEE), “with proper planning, the PCF is able to deliver a near-isochronous service to the stations on the polling list. ... While the PCF is an optional part of the IEEE 802.11 standard, *every station* is required to be able to respond to the operation of the PCF”. That means, access points are not required, but if an access point is present, every station is required to operate in a polled mode. Schulzrinne: Except that APs donʼt actually use it. We have worked extensively in this area recently; see our WCNC paper, for example, for PCF extensions. (http://www.cs.columbia.edu/~hgs/papers/Kawa0503_Improving.pdf) This is just an example, and a rather specialized one; there are many more papers, some cited in that paper, on VoIP capacity for DCF (and a few on PCF). The problem with some PCF-is-better papers is that they assume that the PCF interval and the voice interval are per174 fectly lined up. Nothing ever is, so you get into the situation where the station misses the poll by just a tad, then sends during the contention period (as called for) and then has nothing to send during the next content-free (polling) period, wasting the poll. Also, with silence suppression, about half of the PCF polls are wasted on temporarily silent stations. Reed: As you certainly know, polled networks are NOT slotted Aloha, and are appropriate for voice. In fact, that is exactly what WiMAX is - a polled network. Schulzrinne: Unfortunately, in practice, the voice capacity of PCF is actually lower than for DCF, at least under the standard voice models weʼve used. Reed: So there is no difference resulting from the Medium Access Protocol between WiFi and WiMax regarding its appropriateness for VoIP (SIP, , ...). Where there are differences, it results from overloading the local access capacity. It is trivially easy to overload a 10 Mb/sec network with a single file transfer. And the situation is worse when your “uplink” (as in a home network) is only a few hundred kb/sec, matched to a 10 Mb/s local access network. Schulzrinne: And SIP works fine in networks where the average load on the bottleneck link is 50% or less, and the biggest packets are a small number of milliseconds long on the bottleneck link. The bottleneck link is rarely the wireless access link, but can be the next link upstream. This isnʼt the result of slotted aloha at all, but is instead the result of overloading the upstream link input queues, creating a transient, but slowresolving traffic jam in the buffers on the source side of the upstream link. Nor is it particularly due to wireless vs. wired link reliability. The same problem with voice happens in wired LANs with under provisioned upstream paths, and the same solutions work, but they are not implemented in The COOK Report on Internet May June 2005 the LAN equipment in either case. They must be implemented in the endpoints or the router queues, or both. Reed (in response to Schulzrinne above): Iʼm ok to discover Iʼm wrong, but I didnʼt base my statements only on that book, which was written by longtime chairs of the 802.11 committee, and vetted by implementers at Intersil and other primary implementors. I based them on actual empirical observations carried out by me and a student in exploring the actual MAC layer of installed APs at MIT. I may have even looked at Netgear and Linksys APs at my home in the process. PCF was present in all cases. This was not about VoIP, so I bow to Henningʼs experience with VoIP problems on WiFi. [snip] Roberts: I want to thank both Henning Schulzrinne and David Reed for helping me understand what the problems with 802.11 for VoIP are and why I have not been able to make it work. Clearly there are settings to optimize but even then FTP in the same zone (which we have lots of) makes delay grow quickly. Also the number of phones is limited far more than the “10 Mbps” would suggest. But from this information, perhaps we can control the traffic to make it work, up to the natural limit. Thanks! And on March 1 David Reed: Henning, et al. - I have spent a few days in my spare time doing exactly what I often suggest others do, checking with my own hands and eyes. Executive summary - you are right, and I was wrong. The vast majority of access points I scanned claim that they donʼt implement PCF, so they donʼt support a polling mode, and instead are CSMA/CA. [the work last year had me looking at 802.11 beacon frames, but whether PCF was enabled didnʼt matter, so I guess I misremembered :-( ] Kismet is a quite reasonable tool for examining the protocols implemented by access points, to see who claims to implement the PCF function on cheap access points, as opposed to DCF. When one wants to pursue the engineering truth (rather than the marketing FUD) about networking technologies, it is very useful to do hands-on exploration. (I remember the many demos of “bad quality over VoIP” by marketroids that came down to the fact that most PCs had problematic sound cards and drivers and bad VoIP software that didnʼt understand how to use those devices in low-latency mode). The mystery of networks to users, and sometimes even “experts”, often leads to treating the somewhat inaccessible phenomena on networks as “magic” confusing correlation with causation, especially if one has a prejudice (wireless is unreliable) that makes the explanation plausible and confirms the prejudice. However, a few minutes (or hours) with Ethereal or Kismet often leads to a different conclusion, if you have the eye for it. I strongly suggest being skeptical of all “experts” (even/especially me! :-) - I suspect that no one is more skeptical of what I think to be true than I am, which is why I went back to check myself). I still believe that the bulk of problems with Skype and VoIP over WiFi arise from overloads at the upstream access link, not with WiFi per se. In any case, sorry for misleading you all. Vonage Whinning its Way to Open Access p. 129 Editorʼs Note; Early March saw an example of what will undoubtedly happen again: port blocking used by a small North Carolina CLEC. Vonage asked the FCC to step in and prevent the blocking. On March 3 the FCC ruled in Vonageʼs favor. http://www.internetweek.com/allStories/showArticle. jhtml?articleID=60405234 “The Federal Communications Commission announced Thursday that it had reached a $15,000 consent decree with Madison River Communications, a Mebane, North Carolina service provider that calls itself the “17th largest phone company in the US,” with “234,204 connections in ser175 vice.” According to the FCC, Madison River pledged to “refrain from blocking VOIP traffic and ensure that such blocking will not recur.” This touched off a lot of discussion. I am publishing only a small part of the discussion because it goes too deep into the American regulatory arcanae that in the view of some been responsible for the failure of broadband infrastructure in the US to keep up with the rest of the world. Matt Wenger: Vonageʼs response is perfectly indicative of what is wrong in Telecom. The infrastructure providers want to have a monopoly on services (we own the customer.) The pure service plays want no-cost access to infrastructure (we own the customer and you, Mr infrastructure provider, own the duncecap) The customer wants a wide variety of affordable services with as little hassle as possible. This fleeting moment where the rules of engagement have fallen behind the ability of the technology (read: loop-holes) is not sustainable. It is acceptable for Vonage et al to take advantage of it while it “fleets”. It is childish to whine about it when it “flows”. We need a sustainable model and what weʼve got now, Mr. Vonage, is not it. If you ainʼt part of the solution... p. 130 Wenger: Wha-ha? Sure you can. You can do anything you want in a deregulated environment where you own the asset. Watch what happens when these guys build fiber. Have you seen the FCC ruling? Now you are right, we are in a weird grey space of non-re-un-reregulated telcom, but on a basic principle if I own the asset, and I am a private company, I can do what I want. This is called de-regulation and it is where we are going, like it or not. But letʼs drop the ILEC case - Iʼd love to bully the ILEC as much as the next beanpole with a penchant for regulation/public floggings. What if the company that Vonage is complaining about is actually a CLEC? Would this change your argument? Would you say, oh well no, I guess they can block? What if it was the cable COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA company (which my badly informed, usually delusional contacts tell me this is the case)? What if it was the local wireless ISP? What if it was all three and they were all competing in the same geographic area? Are you arguing that no company with an asset should be able to control that asset (and protect their investment) or are you just arguing that the assortment of ILECs canʼt have that right? Schulzrinne: This illustrates the need to separate out the rate-of-return, natural monopoly parts (fiber, coax, copper, licensed spectrum) from the competitive aspects of service provision. Wenger: Now youʼre talking my language. I couldnʼt have said it better myself. You would have to look a ways to find a greater believer in (or advocate for) pure Open Access than I. And youʼre right. Thatʼs exactly what this illustrates. But when you do that, how do cover the costs of the Infrastructure? Snip Basically Henning, I think we hold the same belief. The ideal would be separation of infrastructure from services. What I think we disagree on is how the dollars might flow to sustain such an ideal. Iʼd love to hear a different model of sustainability for an open system if you have one. While weʼve had a lot of success (after a lot of failure) with our models, there is always room for improvement or even radical innovation. How to Explain Why Skype Works Better? p. 133 Retzer: How do you explain that Skype simply seems to work better than H.323 and SIP? Neither of those would, I think have met the Sherpa test as well or the link reported yesterday WIMAX local - - Internet - PSTN - T-mobile even assuming they found their way to the PSTN. The quality probably would have been poor. I gave up on Vonage locally last year because quality was so poor over my Comcast account. Skype works great, at least on the limited tests Iʼve tried. Iʼve similarly found h.323 calls need to be engineered to work well but Skype , as they say “just works.” Not being able to look into how Skype works or how the flows actually go (which is why I asked if you or anyone else has tried to analyze flows), one hypothesis I have is that is trading latency for jitter and bit loss - accepting delays to otherwise improve the quality, maybe using P2P to cache calls. Obviously, codec choices may also have a huge bearing. Your comments, below seem to support this notion to some degree. Have you done any actual analysis of Skype? Thanks Forster: Better execution. Architecture counts for about 10% of a solution; execution & operations count for about 90%. In this case the area of great execution is that Skype apparently downshifts to lower bandwidth and/or larger packet sizes quite nicely by itself. Thereʼs no question that the codec in the phone Cullen sent could do a decent job on a 64kbps link, but neither Cullen nor I had the time to fiddle around with the parameters. As Cullen mentioned earlier, Skype licensed a codec and packetization engine from some outfit and theyʼve done a great job with it. Iʼm happy to give Kudos to Skype for their very nice job, but letʼs not confuse a great product with other architectural and protocol issues. Skype wonʼt be the last really great voice software product. Reed: Hear, hear! Jim Forster - you nailed it with that line. But the folks who think they can sit this out in comfortable enterprise niche markets, selling FUD to the CIOʼs about requirements for perfect QoS, -- in order to win closed, proprietary, incompatible bids that interoperate merely as crappy interconnects with the PSTN -- those guys are toast in the long run. Remember that Cullinane (king of the database market) *deigned* to do a crippled PC version of their software that was carefully designed not to cannibalize the customers for their main product. Where are they now? Wall St. “loved” their embrace of the PC market, by the way, and 176 thought they were really “cutting edge”. Shows you how smart it is to listen to the investment bankers on these things. Everybody loves to suck up to the selfdefined brilliant CEOs who “bless” the new technology with their charismatic narcissism. (Leaders are narcissistic by definition, which is what gives them charisma, often a very good thing in a leader, so thatʼs not intended as a criticism; rather I am emphasizing why it leads them to be somewhat blind and insulated by sycophants from really disruptive, worrisome changes) I doubt that Skype thinks it has won the market. I think they are still scared, which is a much better attitude than those who think SIP is merely an incremental way to do the same old stuff over IP. p. 134 The Best days of Voice-over-IP Telephone Service May Already Have Passed By Robert X. Cringely These are heady days for Voice-over-IP (VoIP) phone services. From Vonage to Packet8 to Skype and a hundred more besides, several million people around the world are enjoying really cheap phone calls that are carried primarily over the Internet. But that fun may be diminishing soon because the big Internet service providers, which is to say the big telephone and cable TV companies, are about to start taking back that third-party VoIP traffic, leaving Vonage and the others at a distinct disadvantage. [Snip] The other big broadband Internet provider is typically your local cable TV company. Not sharing the values of phone companies, they ignored VoIP, too, until they realized that it could become a new source of revenue. With so many things available for free on the net, finding a service people are willing to actually pay for is thrilling to companies like Comcast. Now they just want to make sure that the VoIP service you use is THEIR VoIP service. The trick for phone companies and cable companies alike is to hurt the VoIP up- The COOK Report on Internet May June 2005 starts without incurring the wrath of Congress, the FCC, or any other regulator. They have to be sneaky. Hereʼs how they plan to cripple the Vonages and ʻs, according to friends of mine who have spent 20+ years in engineering positions at telephone companies, cable companies and internet service providers. As the phone and cable companies begin offering their own VoIP services in real volume, they plan to “tag” their own VoIP packets so that at least within their own networks, their VoIP service will have COS (Class of Service) assignments with their routers, switches, etc. They also plan on implementing distinct Virtual Local Area Networks (VLANs) for the tagged packets. p. 135 Menard: Frank, If your message had come from anyone else, I would have said that your message had been written by someone who doesnʼt understand the challenges of competitive entry. Evolution is the only way in telecom because regulators will not let revolution happen. There are too many jobs at stake. Reed: Ever the optimist, I just refuse to cede such power to the regulators. Of course thatʼs what the *tendency* is, Francois. But the regulators can be beaten. Just for example: twice in my life the US Postal Service tried to get Congress to block “electronic mail” because they were supposed to have a government monopoly (including being the largest jobs program in the United States!) according to the law. But Are P2P Voice Applications Blockable in the Same Way as Vonage or Lingo? p. 136 Meanwhile I have a technical question. we have seen it mentioned that the cable plant can use docsis to descriminate against Vonage, Lingo and the like. Would really appreciate it if one of our more technically minded folk would comment on the dangers here for VoIP traffic. What is the common denominator here of susceptibility if anything? SIP? Something else? Port assignment? The FCC doesnʼt like port discrimination but unless Brand x is upheld it will make no different because the cable cos are regulation free. (Without going back to the article) was he saying that DSL telco folk could do the same thing? And finally and most important what about the P2P services like ? How susceptible are they to traffic shaping and prioritization services? Has anyone tried to strangle a p2p service yet? They have gone after them legally but could the incumbents kill in the last mile? I would have thought not but i certainly donʼt know the technical detail the way many of you do. Reed: You can *try* to block , Vonage, Lingo, Bittorrent, etc. at the DOCSIS layer, but the fundamental issue is “recognition”. Recognizing a specific end-to-end application can be done by looking for “signatures” specific to that application. To do so, you have to reverse-engineer the application, and distinguish it from other applications you dare not block (because they would cause customers to switch to another provider, at least in a competitive market). Thus, I could probably come up with a scheme that blocks Skype, but if I and many other providers start blocking enough to annoy Skype, it can release a new version that doesnʼt have that signature. Ultimately this “arms race” can be won by the ISP only if a) it can keep its customers from switching to a competitor that does not block (and in the US, oligopoly attempts to organize competitors violates antitrust laws, so you canʼt do this by forming a club of thieves to conspire to define a blocking strategy across all competitors in a market), 177 or b) it can find a class of behavior that it can block that is acknowledged to be “bad” by all competitors and the government. Linking the blocking to that class is possible. For example, one could enforce a rule that every 10 seconds the access port refuses to send ANY packets for 0.5 second. This would kill all isochronous uplink traffic, but allow all of the couch potato traffic including streaming video to continue to work. You still have to convince Congress that a conspiracy by access providers to deploy such a scheme is worth not enforcing antitrust laws against. Assuming one could get the government to agree that preserving the right of the carrier to block isochronous uplink traffic will prevent, say, Child Pornography or Terrorism, youʼve got a deal with any Congressman willing to take your PAC money. Thereʼs room for creativity here - thereʼs no lack of willing technologists out there who will creatively argue that there is “risk to the network” if certain applications are allowed to flourish - remember Carterfone and Hushaphone (and remember that in Hushaphone, the courts were persuaded that a piece of plastic on the handset threatened the entire network). Summary Conclusion p. 143 Editorʼs Note: One of the enduring lessons that I am carrying away from a three month immersion in SKYPE SIP and VoIP is that while it may yet be a long time before voice leaves the PSTN, that the richness of opportunities to communicate is becoming so stunning that the very multitude of new opportunities will lead to difficulties. It starts from the very scarcity model of the phone company where we communicate one on one at the sufferance of the commanders of the “intelligent network” and where our identities as phone numbers are on view to the world unless we pay to keep them hidden. What is emerging is a much different model where if we donʼt talk more direct control over our identity and reachability we will be overwhelmed. The new tools can be used in many new COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA ways as the discussion with which we shall close this symposium shows. What is uncertain is how far and how fast this perceived need for identity management will spread. The user communities involved with tools like Skype will need to work out ways of using the tools that are acceptable to others. Skype had better listen carefully ti users here and incorporate with in its product the ability for users to control the extent to which they may be reached and conditions under which they may be reached with these tools. And above all Skype had better look at engineering barricades against spam. This is early adopter leading edge experience to which the Skype community, the handset makes and the phone companies at large need to pay careful attention. COOK Report: On March 11, Jame Enck posted a short essay he called Skam on his blog http://eurotelcoblog.blogspot. com/ Iʼm going to coin a phrase here, I think. Anyway, Skam = Skype spam. A friend received the following message yesterday (typos remain): Dear friend, I represent a consulting company from Latvia. The company is one of oldest in renewed Latvia. I have a proposal concerning legaly reducing income taxes. You would be supprised about the results. Agents interesets are appreciated. p. 144 Henshall: The opportunity exists for Skype to insert another level of chat control. This option would automatically refuse chat invites where more than x people are involved. Or said similarly, “limit” my chat participation to chats containing 4 or less. The mega group chats can then be always “blocked” or put into a call/chat list context that doesnʼt pop-up/ring buzzers etc. This would be useful. If you have ever had a group chat going while being on a SkypeCall and havenʼt turned off the message “whistle” notification you will know what I mean. Creating two chat types would add new utility. There are pervasive chats that many want to leave open, similarly the head of a small company may start a group chat to provide an update. Traditionally this may have gone by e-mail, some are even broadcast phone messages. Sending it via Skype means you will get it where ever you are logged on. Thatʼs interesting in a mobile, multi-device world. It seems like it may follow you better than e-mail and, along with presence, be more timely. The Skype mega chat is similar to an e-mail list. However, to give it the priority of a one to one chat is wrong. One to one chats are invasive, less so than a call, and more so than just leaving a voice message. p. 145-46 Davis: Stuart, I remain to be convinced of your premise (RE your remarks on Skype vs Aastra), that being that you know the complete story on “productivity” and “changed work practices”, “real opportunity”, what will “advantage” an unlimited set of users. If the butter on your bread is spread by the knife of “presence”, “always-on availability”, your bluetooth headset to softphone, then more power to you. Such is a different statement than saying what is best or better for “everyone.” In my world, such is demanded at the level of the first line supervisor. Span of control and scope of work require increasingly filtered communications “presence”, mitigated by priorities, up the chain and across scope focii. Human transactions are touted in “popmanagement” and “pop-psych” literature as equivalent to worker “empowerment, but only for the dispossessed. Human transactions are a variable overhead cost that must be run through the calculus of cost/benefit analysis once human organizations reach a size of more than 50. Just as industrial design and engineering has deployed statistical analysis on product quality to find that the overwhelming preponderance of error was at the human-to-line and human-to-product interaction nexus, so also is it for interhuman communication. Ninety percent of human speech is not about the content expressed but about all 178 sorts of other primate agendas, including, but not limited to, posturing for rank, contentiousness, turf, “blue sky” attention seeking, scape-goating, and just pure “keep alives.” Those are calculated on the cost side of the cost/benefit calculus. We develop electronic work flow and electronic document sharing to eliminate those long and personnel cost expensive “reporting” and “project management” conference calls. The margin pressure in the global economy just wonʼt afford them. Henshall: For me, the future is about “Choice of Tools.” As we move from traditional labor and traditional life long job skills and employer homing, it will be up to the worker to select the Tools and how and when to use them to set marketability. More choices and the ability to choose among tools are what “empowerment” means. More choices do not make for better decisions, of course, An the market is the ultimate judge. Savage: Let me de-lurk here for a brief observation. The underlying assumption of 20th century telephony and later ʻnet connectivity is that ubiquity is either the norm or the desired goal. [Insert historical observation/rant about Bell System propaganda value of universal service here.] In fact I suspect that most people have little interest in connecting to the vast majority of others on a “ubiquitous” network. Moreover, interests in connecting are asymetrical. The group of parents on my kidsʼ soccer teams need to be able to call each other. I want to be able to call tech support for the products I buy but I do not want to hear from the companyʼs telemarketers. Etc. My not-terribly-well-formed hypothesis is that if establishing connectivity to selected groups of people were easy and cheap, what would happen is the creation of islands of open connectivity linked to each other and/or the world by non-transparent “gateways.” Think AIM buddy lists. Once you are vouched for by someone in the group, youʼre OK, but until then youʼre out. The COOK Report on Internet May June 2005 This absolutely interferes with ubiquitous connectivity. My suggestion is that this is not a bad thing. To the contrary, it is perfectly rational for people to value connectivity to certain groups more than others, to value inbound versus outbound connectivity differently, etc. Skype et al. are permitting a certain natural experiment of this phenomenon in a way that the ubiquity model, imposed via the command-and-control Bell System, did not. Skype-In Looks to Be an Important Intersection Between Skype and SIP p. 147 Editorʼs Note: Skype-In announced on March 11 and Skpye-Out now a million strong rests on an agreement with i-Basis to use SIP to interconnect calls with the PSTN. Shockey: (Please see Stuart Henshalls Interesting blog entry) http://www.skypejournal.com/blog/ archives/2005/03/skype_strategy. php#more Henshall: What should Skypeʼs strategy for SIP be? With the beta launch of SkypeIn many Skypers are getting a “traditional” phone number. That number is also a SIP number. The Skype intereconnect (see iBasis below) to the PSTN required working with partners that could “connect” Skype to the PSTN. Thus all SkypeOut and SkypeIn calls use SIP to connect. This is a question piece, looking at whether Skype should make SIP numbers available to SkypeIn buyers. Then connecting to any SIP phone from Skype is free. (See also Skype and SIP?) Consider -Should Skype offer Skypers their SIP number at a small price premium? This would immediately separate them from almost everyone else outside FWD and PhoneGaim. Vonageʼs ATA boxes are locked, CallVantage the same, etc. Even most of the “new” softphone players (TelTel, Teleo, Damaka) are “locked” even when they claim SIP. What are the strategic implications for Skype and competitors if Skype opens up SIP? As a user I could use Skype “supported” SIP features. That may mean I can ring a Wi-Fi SIP phone device, I may find it hard to scroll my buddy list or obtain the same voice quality as I get with the Skype experience. Would this allow me to receive “securely” (I donʼt buy the security argument BTW) my SkypeCalls on a corporate SIP device or SIP softphone without upsetting my IT department? If so this would be pretty cool. At the moment unless I have call forwarding, (or Vonage already at home) I canʼt enable my home landline to ring the office. Wouldnʼt SIP enable you to ring your Skype almost anywhere? While the call quality would be inferior... the sound would always be better on a Skype soft client or via a Skype enabled hardware device (the approval and integration process is real important). What if Skype just offered SIP numbers with SkypeIn? Would that make them quickly the largest SIP community and deployment? Could Skype then “shape” the direction of SIP and thus telephony. What happens if Skype introduces their own version of SIP (Microsoft vs Sun on Java)? p. 148-149 The differentiator in user value therefore is in the unwanted connections. A totally open SIP solution with a “naive” ENUM-type call routing solution will result in the mother of all voice spam festivals. This is likely to swamp any positive benefit from increased reach of addressable endpoints. Richard is already working on adding pointer data to ENUM as the first steps to creating identity meta-data you can use to filter callers; there remains a lot 179 of work to do. Note that the alternative -- Carrier ENUM -- is again a closed “members only” system. Skypeʼs closure is therefore a feature, not a bug, in the absence of a scaleable public identity system to complement an open protocol. If you look at some of Stuart Henshall or James Enckʼs recent postings, you will see how important are limiting communications and connectivity. The features they seek are to *not* participate in large group chats, or *not* to receive long voicemails. SIP and ENUM attempt to import the best of SMTP, HTTP and DNS. Unfortunately, they also import the worst aspects of those, and without regard to the fundamental differences between media they carry. Unlike e-mail, you canʼt use the message content to filter voice communication. The interruption -- and cost of the unwanted call -- happens up-front. SIP and ENUM cannot survive as currently constituted by copying the open model of e-mail. Today they can only prosper as hidden components of closed systems. In the past the value of telephony came from connecting people over scare connectivity. Now the “connecting” effort over and above the packet data costs are negligible. In the future value will come from *inhibiting* connections over abundant connectivity. The paradox of the paradox of the best network! (See http://www.telepocalypse.net/archives/000407.html). You make more money from stupider networks by throwing stuff away intelligently. COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Introduction and Executive Summary The Telephone Becomes Software & the Phone a Computer Skype Examined as Virally Spreading 2nd Gen of VoIP While SIP & World of Enterprise VoIP Live in Very Different Universe April 3, 2005 Ewing, NJ The telephone has become software. In 2003, VoIP was a bucket of bits adapted to a special SIPendowed handset or “Ethernet phone” that cost between $200 and $600 and had to be connected via a complicated interface to an early island of connectivity like Free World Dial Up. In 2005 VoIP is viral and software. It is a program called Skype downloadable from the Internet. Other such programs will follow. Convergence Arrives at Level of the End User Device Skype is a free software phone that “just works.” And, as our experts point out, Skype is not necessarily the be-all and end-all of a softphone. Better products will likely appear, and, whatever happens, the business model appears to be profoundly disruptive. At the very end of March Skype CEO Niklas Zennstrom published by Mathaba on March 29, 2005, said, “The telephone is becoming a computer. See my telephone, I have Skype on a mobile phone with a 200 MHz processor and an operating system, and it’s connected to the Internet with different types of radio access, whether its GPRS, WiFi, 3G or WiMax, it uses different radio technologies. Then telephony becomes a software application that just lives on the network. Skype is that software application, there will certainly be others...” [Source is Skype Journal, Skype on my Linksys Router, April 2, 2005.] Lingo and Vonage are based on hardware devices that must be shipped to customers and added to their networks. Both have the old advertising, network maintenance, billing and customer acquisition costs of the phone company business model. Customer acquisition costs are likely well over $100 per customer for services that bill at about 20 to 25 dollars a month. Skype customer acquisition costs are more like a penny a customer for a service where over a million people have purchased the minimum 10 Euro Skype-Out service and many many more merely use the service without ever leaving the Internet. Two broadband connected people anywhere in the world can download the program, register an ID in the Skype directory and be speaking to each other with better than PSTN quality sound in less than 30 minutes. The user’s computer needs only a speaker and any microphone. Because Skype is built around Instant messaging including file transfer capability, users having their hands free can use the internet connected capability to exchange text, pictures, URLs - in short any form of data in real time. The possibilities created for productive collaboration are enormous. The channel of voice communication with a handset and receiver formerly held to the ear goes, in an instant, from very narrow to very wide. The nature of the interaction frequently changes. Fore example switch from headset to speakers and the conversation becomes more social to those present. And for those with the computer and broadband, there is zero cost paid to the phone company. Zero cost - for high quality collaboration anywhere in the world that broadband exists. Zero incremental cost also creates new oppor180 tunities for use such as "squawk boxes" where the "call" is always-on. It is a rather startling change that is best experienced first hand rather than described in words alone. Furthermore given Skype’s peer-topeer architecture, broadband providers - phone companies and cable companies and potential competitors of Skype will find it almost impossible to discriminate against it. Its packets, encrypted will be harder for the provider to identify without the expenditure of disproportionate effort. The IP address of the authentication server could be blocked as could the IP addresses of known supernodes. If that were done, Skype could fight back with its own technology changes and enter a war of escalation that it would be unlikely to loose. Not so for last year’s generation of Ethernet devices (Vonage, Lingo, Packet 8 and others) that link a customer’s handset through the Ethernet telephone adaptor's MAC address to an authentication database and compress and packetize the audio sending it through the network along with email, web, and other easily-identifiable data streams. Should it decide to do so, the telco cable-co duopoly in the US can favor its own higher priced VoIP services and give the so-called parasitic services whatever is left on a best effort traffic basis. Robert Cringley explains how on p. 133 above. Despite the FCC’s favorable ruling in March on behalf of Vonage, if the telco and cable cos, with VoIP packages of their own to sell, take the approached described in Cringley's article (see p. 133 and following discussion), they will very likely kill these non P2P “parasitic” VoIP services. Imposition of PSTN access charges by the FCC The COOK Report on Internet May June 2005 is another source of their possible demise. For these reasons and in my opinion, VC's who invest further in Vonage may find themselves on very thin ice. Since Skype is P2P and uses the CPU and drives of the more than two million users who are now routinely simultaneously attached to its network, Skype will, for the most part, be able to avoid the centralized control functions exerted at the cable head ends or the telco CO. While it does have a centralized authentication server, reportedly in Denmark, I have yet to find someone who asserts that this server could be Skype’s Achilles heel. If one managed to bring down Skype, one will find that there are even more decentralized softphone programs by Popular Telephony and Nimcat Networks waiting in the wings. These programs claim to be "serverless" and to give their users the capability to install them in a federated fashion where user groups can define limits of inclusion and install bridges that will connect islands to each other. Doing this would create another decentralized architecture that the telco and cable cos would find hard to root out. Meanwhile, running on the major desktop platforms and in more and more cell phones, Skype is spreading virally. [See the May 19 2003 “A Viral Communications Architecture” paper at http:// dl.media.mit.edu/viral/viral.pdf. We discuss this on page seven of this issue.] On Sunday April 3, Richard Stastny, a major contributor to this issue, placed the other “bookend” around the growth of Skype asking "Is Skype reaching The Tipping Point?" http://voipandenum. blogspot.com/2005/04/is-skype-reaching-tipping-point.html "Skype is currently growing epidemic, so one may step back and look-up again “The Three Rules of Epidemics” given in the bestseller from 2000 “The Tipping Point” . . . . The book analyzes why ideas, trends and products seemingly out of nowhere suddenly cross a threshold, tip and spread like wildfire." I agree with Richard. Skype is a major development. With its API opened, Skype is gaining third party support that permits a wide range of communication functions. It looks like it will be soon merging voice communication on cell phones and with Wi-Fi. As long as the bandwidth is adequate (at least 64 kilobits or better) Skype can be used on mobile platforms with which it is compatible. If you can browse the web on your cell phone, increasingly you will be able to Skype from it. Skype has signed an alliance in the UK that will open wi-fi hotspots to people who wish to use Skype and do so without charge. For business people the savings involved in avoiding cell phone connect charges are very large. Its Skype-Out service that permits Skype calls to the PSTN for a small per-minute fee passed a million users for the first time in late March. Dave Hughes and I began to use Skype to call Nepal in late November. Then on December 1st, Hughes began a symposium mail list discussion thread that led to this May June COOK Report issue when David Reed grabbed my attention a week later with the statement that Skype was on its way to becoming the globally dominant “Wintel platform of Voice over IP.” SIP, Reed continued, had failed to become the signaling protocol glue that was going to tie all manner of different VoIP hardware devices together because the SIP developers were too beholden to the phone companies. Reed asserted that the "way to succeed in business is to pick the best customers, and delight them. And the crucial caveat - the best customers are not the ones who always buy anything you sell - those are *your* best customers, not *the* best customers. The best customers are the ones who will teach you what you should be selling. The following is how it applies here: SIP’s vendors have defined their customers to be phone companies. Skype has defined its customers to be people who live a communications-centered life. 181 It’s impossible to delight a phone company with voice over the Internet. The people who live a communicationscentered life will teach you what really matters. Those people are *not* happy customers of the phone company. It’s still possible to beat Skype with SIP, but the current SIP vendors (such as XTen) have no clue whatsoever! To win, you have to delight some customers, not participate in an illusory “market” for “technology” Reed concluded." Seeing this as a provocative challenge, I invited Richard Stastny, Richard Shockey, Henning Schulzrinne, and Cullen Jennings to join the discussion list and later added Martin Geddes (Telepocalypse), James Enck (Eurotelcoblog), and Stuart Henshall (Skype Journal). The result was a fascinating barrage of information that made the material for this issue too encyclopedic. (Yes - even I have reached my limits - feeling like I am constantly chasing the multiplying broomsticks of the Sorcerer’s Apprentice in putting the final details for this issue together.) The pace of change with Skype between the beginning of December and the first days of April is amazing. Skype has had a significant announcement of new features and capabilities or a new alliance with the likes of Motorola and others almost every week. For example, a new update for Skype Windows on March 23 (1.2) is features expanded address book capabilities and integration from other address book sources. http://skype.com/company/news/2005/ v1_2.html But Donʼt Forget SIP Meanwhile Henning Schulzrinne and Cullen Jennings provide, in this issue, admirable defenses of SIP which, as a signaling protocol, is burdened with the need to interoperate with all kinds of different devices. But Martin Geddes made some very worthwhile comments on SIP as he read a draft of this summary: "SIP is not very important to understanding the future of VoIP. Talking SIP versus COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Skype is a bit like trying to analyze ecommerce by reading about HTTP." [Editor: The remainder of Martin's remarks are found on page 184 below.] way that nothing in more than 100 years of wire line voice communication has ever had before. Free, unencumbered, and viral it will give the enterprise fits! Whether SIP has truly failed or not depends on one’s perspective. Even Skype depends on SIP for PSTN gateway signaling for Skype-In and Skype-Out service. SIP has been burdened by many requirements for making a wide variety of hardware communicate with each other under widely varying and complex conditions. Interoperability demands have yielded results of protocol complexity. However, as Melissa Davis points out, enterprises - especially the larger ones - have information and data security requirements that are legal and fiduciary and cannot therefore be lightly dismissed. Because of these requirements, Skype – as it now stands - is unwelcome in a corporate network. Skype, as a proprietary application installable on a corporate desktop and one that uses the open port 80 connection by which corporate employees can surf the web to send encrypted data streams through the corporate firewall, is a security risk. In effect, Skype breaches the Maginot line that is the central definition of corporate security. Here the penetration is really unique since it goes from the inside out. Coping with this demands quite a role reversal for the security managers who are trained to protect against incursions that come from the outside world. In other words: Where is the appropriate place for standards? You have to decide in advance who your customer is and realize that what one customer set wants the other may hate. What then do you do? Do you give away something proprietary like Skype and capture huge user base? Or, if you aren’t going to give it away, do you need something like SIP to make it interoperate? There is great deal of truth that vendors develop technology that their big industrial customers will like. But technology that pleases these big companies is not necessarily technology that pleases end users. CIisco’S 40 gig router being a case in point - the carrier customers want it but do they represent a market with enough longevity to justify Cisco’s investment? In the VoIP world David Reed recommends: Ride the commoditization curve of Instant Messaging instead of trying for monopoly rents. But I ask: How do you create wealth through commoditization? Enabling the power of what one person can do to move to the edge? And then develop loose networks of on-call specialists? Skype and the Enterprise What I am seeing with Skype that is new and different is that, since it is a soft phone, it becomes entirely freed from the old legacy infrastructure of physical network boundaries and devices from the switches in the network center to the PBXs and the like at the edges. It really does have a Greenfield to grow from in a While the ordinary telephone facilitates communication in a way that is congenial to corporate hierarchy, Skype with it’s buddy lists, and its ability to conference in five callers with a few mouse clicks and its ability to keep audio channels to people on the outside open all day long represents a huge threat to both the real and perceived need to guard corporate information. For an eloquent description of what is being described, readers are invited to turn to pages 105 through 107 where Melissa and David Reed (two people with very different perspectives) debate and, in the end, arrive at a shared understanding of the issues. The enterprise aspect of all this is a major issue. There is substantial money in enterprise voice. I predict that Skypelike softphones will penetrate the corporation. The advantages of collaboration by freeing professionals to use to use these collaboration tools that are so intimately bound up with their PCs and that leverage Instant Messaging with retainable reviewable archives of every use are too great to be kept forever at bay except by the most fortress like institutions. The 182 major unknown is how quickly the penetration will occur. The discussion in this issue explores several scenarios. There is some evidence that some enterprises may be exploring customized versions of Skype. Skype may itself bring out a version designed to alleviate security concerns. There are also emerging soft phone alternatives from Nimcat Networks, Popular Telephony and before long others that will offer voice communication integrated with broadband networks rather than with the old wireline PSTN. Skype as a Second Generation of VoIP I am concluding that the profound story here in “VoIP land” is not Vonage and Cisco's "stuff," or SIP, or IP PBXs etc - this "stuff", one way or another is tied to hardware and not surprisingly to the phone system and its related hardware. What I am increasing certain of is that Skype represents a second generation of VoIP that will have a far more profound impact than the first generation. The phone as software does not begin nor end with Skype. Analysts had better study Google’s emerging local services. Google is building a directory of local businesses that is keyed to maps that are far superior to anything done so far by Map Qwest. It is likely that Google will begin to offer these businesses its own Skype-like soft phone services. Do a local merchant search. Click on the phone icon that Google returns for a business and open a voice connection with the merchant. It is doable now. We have reports that Google is testing it in the United Kingdom The reaction of many people is going to be” “Why would I use my computer to talk to someone rather than the telephone?” Let me say only that this immersion in Skype and softphones has left me with the conviction that, when people try it, they will find the experience to be so comfortable and productive that they would go back to the old way of doing things only with great reluctance. The COOK Report on Internet May June 2005 A New View of Broadband The past three months has encouraged me to see broadband in a wholly new way. Broadband is like a fertilized field on which new crops and creativity will be grown. While the user-controlled or switchcontrolled lightwaves we looked at over the past two months are tied to fiber and switches and huge band width and therefore relevant to research entities and enterprises, this equally profound instant voice "stuff" is not dependent on its users having access to their own special dedicated hardware and big infrastructure and therefore it is NOT dependent on getting a start in enterprises that can give its employees the dedicated layer one and layer two switches as is required for User Controlled Light Paths. Instead, this "stuff" (Skype), free and viral, is tied only to one thing. To BROADBAND and its underlying infrastructure. It could care less whether it rides over DSL or cable modem. It does both with equal ease. And now it is being freed to ride on mobile. Namely on Wi-fi and cell phones. Skype certainly depends on all the existing layers of the protocol stack. In that sense there is nothing magical about it. In depending on broadband it has to depend on the infrastructure that enables broadband. However, while end user premises and connections would need some serious upgrades to run end user-controlled lightwaves, these same end user connections, if they deliver broadband, can deliver Skype without any changes to the underlying network infrastructure. Skype represents a hub to which many different means of communication can be tied, if you have broadband. The emphasis here (with Skype) is that communication is portable and tied to the individual and not to a phone company or a phone network. The lesson is that there are many different pieces here that can be tied together in fascinating ways and, when we are done, anyone who tries to control its users or keep them within the bounds of their network - looses. Years ago SIP had a glimpse of the future with “presence” and a series of imaginative activities that could flow from it. But, through no fault of its own, SIP, has lost by being forced to fit what it did on top of the enterprise and PSTN voice networks, and by being forced to be interoperable with legacy infrastructure. Wrong. Wrong -- in that the only thing that you have to fit voice as a bucket of bits into is packets that ride on TCP/IP or maybe even UDP broadband data pipes. In such an environment, broadband is a foundation for everything else. It is so in a way that I believe is not yet well understood. Skype will move forward in the rest of the world whether the enterprise likes it or not, and the applications that will grow up around Skype will give broadband another huge push that just email and the web cannot. But tragically, this newly productive world is not so likely to flourish in the US because of roadblocks of ideology and ignorance of key elements of technology. Having won at the FCC, the ILECs have a full court press going in the state legislatures. The outcome may be grim. Broadband is infrastructure. It is not frivolous entertainment designed to be delivered by private enterprise for the benefit of the stockholders. Wireless muni-nets are under siege while advanced fiber transmission capabilities are increasing beyond what could be imagined a year ago. One thousand channel wave length multiplexing in a single optical fiber was demonstrated in early March over a 126 kilometer distance in Japan. See http://www.ntt.co.jp/ news/news05e/0503/050308.html A ubiquitous bandwidth utopia, is achievable rather than the current scarcity model. But nothing will be gained until we rediscover that governments do exist to maintain water, sewer, electricity and roadways and the people of a local community deserve sufficient respect from their local politicians to be allowed to invest in and build their own economic foundation rather than being kept subject to the duopoly of the ILEC and Cable Co. From "Is Skype Reaching the Tipping Point?" by Richard Stastny This week (April 3) I was attending the 3GPP/TISPAN/ATIS Workshop on IMS-over-Fixed-Access, basically a meeting of telcos and mobile operators trying to set-up a competition to what I described above. The plans they have are very complicated, mainly because they want to keep up their interconnect regime and termination charges. Since you cannot sell or explain this to a customer, they think they can sell this to a customer by offering a guarantied QoS, even if they still have no idea how to do this. Nobody is asking if the customer is willing to pay for this. [Editor's SNIP] The only asset they really have is the usage of the SIM-card = trusted Identity. Therefore, they should forget anything else and set-up immediately an IMS-light based on SIP using SIM-cards for authentication and roaming. Done. This is what IMS separates from Skype, nothing else. The SIM card is the only sticky thing in a mobile phone. Many of these [telco] people do not even know the enemy. They do not know about or even use Skype. Why? Because they have company-managed laptops and cannot install their own software. [Editor: this is the conclusion of http://voipandenum.blogspot.com/2005/04/is-skype-reaching-tipping-point.html.] 183 COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA Subscription Rates The COOK Report on Internet COOK Network Consultants 431 Greenway Ave Ewing. NJ 08618, USA For definite information please see http://www. cookreport.com/subscriptions.shtml Martin Geddes on the Misinterpretation of SIP in the Context of VoIP People fail to distinguish the difference between SIP -the-protocol and SIP-the network-on-ip. Think of it as the difference between HTTP and the Web. We also have SIP and the “VoB” (Voice over broadband or open-SIP-connections-on-the-Internet). SIP is fine. “VoB” or SIP based voice connections is a failure. Why? That is because you also need a User Interface (a Netscape browser in the example of HTTP versus the web). Think of Skype s playing a UI role similar to Netscape. You also need a directory function. With SIP this is DNS. With Skype, it is the Skype directory permitting unique user-to-user communication. Finally you need a means of limiting your incoming communication. With Skype this is the “buddy list”. The final need is the Voice Channel. SIP offers only the voice channel. Skype offers all four: 1. User Interface; 2. Directory, 3. connection limiting Buddy List; and 4. Voice Channel. It can be said that SIP solves 25% of the user’s problem and Skype 100%. Two Key Differentiators of Skype are 1. Low friction of the adoption process; many platforms, zero configuration, zero payment, easy UI 2. Better life cycle management of “calls”; presence; IM (“is this a good time to call?”); history 184 Gordon Cook, President COOK Network Consultants 431 Greenway Ave Ewing, NJ 08618, USA Telephone & fax (609) 882-2572 Internet: cook@cookreport.com How to use interactive features of this PDF. Click here with Adobe hand tool. http://www.cookreport.com/features.shtml