Liquid Presentation AFPIF3 Final
Transcription
Liquid Presentation AFPIF3 Final
Bridging the African Internet Introduction – Mike Silber • Head of Legal and Commercial – Liquid Telecom • Board member – ICANN – .za Domain Name Authority – TENET • ISPA (South Africa) – Management Committee Bridging the African Internet • Introduction to Liquid Telecom • Some context • Development of the African Internet • What’s been happening and what needs to happen? • Different types of operators have different needs and strategies • Challenges and Considerations • Our vision of a connected Africa and….. • The Local Content Issue Liquid Telecom Introduction • • • • • • • • • • Founded in 2004 Part of Econet Group Traditionally a satellite and voice operator Began planning fibre in 2008 Launched high speed fibre linking Zimbabwe to South Africa and the rest of the world in 2010 Launched high speed fibre linking Zambia to Zimbabwe in 2011 More than 6000 km of fibre currently live running traffic Further construction projects underway Services include IP transit, National backhaul, International and National MPLS, International high speed Ethernet and legacy STM-X SDH connections Acquired Zimbabwe Online (ZOL) in 2012 Central Africa Fibre Project • Provides a very high capacity optic fibre network using state of the art technology • The 8,500 km fibre network will connect all the major cities and towns in Zimbabwe Zambia, DRC and link to Liquid’s POP in South Africa using the ring, mesh and star topologies • It s a mostly a construction project • Also provides international connectivity to the rest of the world • Connects to the undersea cables Fibre Build Quality • • • • • • • • • Buried Ducted fibre 1m deep Or OPGW on HV Power lines Nodes have High Availability Power Systems Next Generation SDH and DWDM High speeds on all routes Ring protection Backed up with NOC, and field engineering Metro rings in the cities Fibre to the Premises in business and residential districts IP Transit and MPLS Backbone IP Transit and Peering • • • • • • • • Liquid Telecom Backbone – AS30844 We like Peering! peering@liquidtelecom.com Major proportion of Internet transit supply to Zimbabwe, Zambia, Lesotho Over 300 peers, open peering at JINX and LINX and selective private peering elsewhere. High Peering Ratio Also operating AS30969, AS56696, AS37485, AS37374 Other AS numbers are peering locally in Zimbabwe, Zambia, Botswana Members of BINX, ZINX, ZIXP as well as JINX and LINX Liquid Connectivity Map • Connection live to 4 sub sea cables • Redundant routes live and under development • One Network • Automatic protection at Layer 2 or Layer 3 • SLA that cannot be matched in the region Some Context How we got here …. 2001 SAT3 2002 SAT3 2003 SAT3 2004 SAT3 2005 SAT3 2006 SAT3 2007 SAT3 2008 SAT3 2009 Seacom TEAMs 2010 Main OnE EASSy GLO-1 2011 WACS 2012 ACE 2013 SAex Fibre Capacity 2001 2001 SAT3 340 gigabits Fibre Capacity 2009 2009 TEAMs 2009 Seacom 2001 SAT3 1280 gigabits 1280 gigabits 340 gigabits Fibre Capacity 2010 2010 EASSy 4720 gigabits 2010 GLO-1 2010 MaIN OnE 2009 TEAMs 2009 Seacom 2001 SAT3 2500 gigabits 1920 gigabits 1280 gigabits 1280 gigabits 340 gigabits Fibre Capacity 2011 2011 WACS 5120 gigabits 2010 EASSy 4720 gigabits 2010 GLO-1 2010 MaIN OnE 2009 TEAMs 2009 Seacom 2001 SAT3 2500 gigabits 1920 gigabits 1280 gigabits 1280 gigabits 340 gigabits Fibre Capacity 2012 2012 ACE 5120 gigabits 2011 WACS 5120 gigabits 2010 EASSy 4720 gigabits 2010 GLO-1 2010 MaIN OnE 2009 TEAMs 2009 Seacom 2001 SAT3 2500 gigabits 1920 gigabits 1280 gigabits 1280 gigabits 340 gigabits Fibre Capacity 2013 2013 SAex 12800 gigabits 2012 ACE 5120 gigabits 2011 WACS 5120 gigabits 2010 EASSy 4720 gigabits 2010 GLO-1 2010 MaIN OnE 2009 TEAMs 2009 Seacom 2001 SAT3 2500 gigabits 1920 gigabits 1280 gigabits 1280 gigabits 340 gigabits Cost Breakdown Satellite/ Cable Breakdown of broadband cost - using satellite Other costs, 4% Other costs, International 9% connectivity, 13% Last mile, 34% International connectivity, 54% Breakdown of broadband cost - using cable Middle mile, National 4% backhaul, 3% National backhaul, 7% Middle mile, 10% Last mile, 61% Monthly cost of broadband per subscriber (USD) Comparing Operators and Technologies 90 80 70 60 50 40 30 20 10 0 Niche wireless Wireline Mass-market wireless Last mile Middle mile National backhaul International connectivity Other costs UK Data Usage Mobile Broadband 0.24 gigabytes Residential Fixed Broadband 17 gigabytes Monthly cost of broadband per subscriber (USD) The Effect of Penetration on Cost Per Subscriber 30 25 20 15 10 5 0 Low BB penetration Mass-market wireless High BB penetration Last mile Middle mile National backhaul International connectivity Other costs Submarine Cables will Greatly Increase the Affordability of Broadband 40 35 30 25 Other costs International connectivity 20 National backhaul Middle mile 15 Last mile 10 5 0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Monthly Broadband Prices Across Africa Percentage split of packages below 1Mbit/s 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Above USD1000 USD500 - 1000 USD300 - 500 USD200 - 300 USD100 - 200 USD50 - 100 Below USD50 What’s been happening in development of the African Internet? Very generalised…… African Internet 2002 • • • • • • • Many Satellite Trunk Routes Undersea fibres available but often priced similarly or more than satellite Internet hubbing in Europe or North America Very few cross border connections Some countries have local IXP Expensive High Latency African Internet 2012 Artists Impression… • • • • • • • • Mostly sub sea Trunk Routes Multiple sub sea routes and competition leading to lower pricing Internet hubbing mostly in Europe IXP of choice for Africa is LINX Some cross border connections, mostly selling commercial transit to countries with no landing stations Nearly all countries have local IXP. Many of these have participation from local incumbent Telco Cheaper than before Latency to Europe about 50% less, inter Africa Latency still about the same as a direct satellite African Internet 2022? • • • • • • • Fully meshed terrestrial Internet? Less direct connections to subsea cables? Cross border peering will be a reality But who will do the cross border peering? We don’t expect to see a pan African IXP Will the African Internet backbone be ‘owned’ by African operators or traditional tier 1 USA based carriers? Will there be content and traffic to support this? What’s Happened • Sub sea cables built • Competition (not necessarily capacity) has brought about a rapid crash in prices • Supply of sub sea capacity probably now has exceeded current demand • Terrestrial fibre networks being built • Digging takes longer than laying fibre on sea bed • Crossing borders can take almost as long as negotiating international sub sea Consortiums J • Cross border capacity usually used for commercial IP transit from countries with landing stations to landlocked countries, voice and corporate data networks What Needs to Happen • Investment in high speed last mile will now drive growth in subscribers • FTTH, WIMAX, LTE, ADSL • Local loop unbundling would be nice • Services based around high speed content will also drive growth in bandwidth used by subscribers • More investment in terrestrial trunk routes, both national backhaul and cross border will grow by business cases • The GSM networks (not the Internet) are the main drivers of investment in national backhaul. 3G has capability to deliver to millions of subscribers though not always at broadband speeds • Increased Internet Subscribers x more bandwidth per subscribers, makes the cost per Mbps less expensive, drives subscriber growth Different Folks have different Interconnection Strategies! • • • • National incumbent Telco or PTO MNO or Pan African network as part of a larger group Independent MNO or Independent Entrepreneurial ISP Liquid Telecom National Incumbent Telco • • • • • • • • • Often owns significant national copper and fibre network Sometimes but not always is dominant supplier of Internet nationally Has licence and permission to run most types of Telco service Almost certainly will have made significant investment in a sub sea consortium whether it is using that capacity or not May own and have exclusivity on landing station May have connection to borders where it interconnects to similar entity in neighbouring country But its network almost certainly stops at its own border Cross border connection may be used for voice, IPLC half circuit and OSS, maybe to sell Internet transit or buy Internet transit from its neighbour Problems with such connections are lack of SLA, different networks so probably no actual protection, you need to actually buy 2 links to have any guarantee of QoS Pan African Networks • • • • • MNO Group, Academic Network etc Group has sizeable bandwidth needs and national network in a number of countries Countries are not necessarily bordering Will have made substantial investment in subsea capacity at group level Desire is to link up those networks to a Pan African network, aggregating and hubbing the purchased sub sea capacity at strategic landing points Independent ISP or MNO • • • • Entrepreneurial Bandwidth needs below STM1 Or unable to make commitments in jumps of STM1 Needs resilience as its not possible to survive if single homed on one fibre system • Buys IP transit but peers locally • More focussed on last mile and customer acquisition than long distance infrastructure projects • May have business customers needing international private VPNs Liquid Telecom • • • • • • • Building one Network across multiple bordering countries Licenced in those countries Crossing borders Open to JVs and partnerships A “Carrier’s Carrier” Servicing the needs of all different types of operator Diverse products to support enterprise, home user, rural broadband Challenges and issues • • • • • • • Regulatory conditions to allow growth Co-ordination with other government departments Roads, Power, ICT Policy, Customs and Excise (borders) Co-ordination between countries Political instability Construction projects Cross border peerings will be low capacity in many cases, low capacity means N x E1 = too expensive • Unique African challenges The Result • • • • • • Internet wont develop to a plan Different operators will follow their own strategies African Internet will evolve as operators execute their strategies It will end up being affordable to most users Most citizens and all businesses will be users We think it will look something like this…. The Local Content Issue • For our case we see that 5% of total traffic in of Lesotho, Zimbabwe, Zambia is for South Africa • Mostly South Africa News and TV, Internet banking, email/ commerce • Hardly any traffic at all between other neighbors say Zimbabwe and Zambia or Zimbabwe and Botswana • Compare that to international voice where there is a lot • Pan African interest sites (e.g. Big Brother Africa) are usually hosted in USA or Europe The Local Content Issue • • • • • • • We looked at the top 25 Websites in Nigeria, South Africa, Egypt, Ghana, Cote D’Ivoire to see if they are hosted locally Compared to More developed Internet nations South Africa has 6, UK has 6, France has 10 and Brazil 7 Egypt and Cote D’Ivoire have some Ghana and Nigeria None Worth noting that none are particularly content heavy sites and also that Nigeria is creating the most video content of any African Nation USA, Russia, China its mostly local sites Country South Africa South Africa South Africa Top 25 9th 10th 13th Website bidorbuy.co.za news24.com fnb.co.za Locally hosted Yes Yes Yes South Africa South Africa South Africa Ivory Coast Ivory Coast Egypt Egypt Egypt 14th 15th 19th 19th 24th 8th 13th 21st standardbank.co.za absa.co.zw iol.co.za mtn.ci aviso.ci masrawy.com ahram.org.eg yallakora.com Yes Yes Yes yes yes yes yes yes Go and Create Content • There is a gap in the market • Potentially very valuable African websites and applications yet to be created • Someone will create such content and that person could be in this room • Mark Zuckerberg is 28 years old – He lost 2 billion dollars on 21st May this year….. Questions?