VIEWPOINT - Focus on Mainframe Cost Optimization
Transcription
VIEWPOINT - Focus on Mainframe Cost Optimization
A Compilation of Articles by Industry Experts VIEWPOINT Focus on: Mainframe Cost Optimization Focus on: Mainframe Cost Optimization The Mainframe Strikes Back: 5 Reasons to Stay with the Mainframe This publication was created by BMC Software. Business Runs on IT. IT Runs on BMC Software. The Keys to Effective Integration *199997* VOLUME 10 Business thrives when IT runs smarter, faster and stronger. That’s why the most demanding IT organizations in the world rely on BMC Software across distributed, mainframe, virtual and cloud environments. Recognized as the leader in Business Service Management, BMC offers a comprehensive approach and unified platform that helps IT organizations cut cost, reduce risk and drive business profit. For the four fiscal quarters ended December 31, 2010, BMC revenue was approximately $2 billion. Visit www.bmc.com for more information. The Economics of Computing 4 Guidelines for Maximizing the Value of the Mainframe Published by BMC Software Acknowledgments We greatly appreciate the contributions of the following people and companies: Jonathan Adams, BMC Software Bruce Chen, Galaxy Al deMoya, BMC Software Felix Garcia, Thomson Reuters BETA Systems Tim Grieser, IDC Per Johansson, Volvo Information Technology Bill Moran, Ptak, Noel & Associates Nick Pachnos, BMC Software John McKenny, BMC Software Dr. Howard A. Rubin, Rubin Worldwide Dr. John Shedletsky, IBM Lilian Vieira, CSC BRASIL Editor-in-Chief: Elaine Korn Executive Advisors: Bill Emmett, Roy Ritthaler Technical Editors: Nick Pachnos, Robin Reddick Senior Editor: Lea Anne Bantsari Editorial Team: Janice Brown, Linda Donovan, Kurt Milne, Karen Nichols, Heather O’Neill Technical Reviewers: April Hickel, Jay Lipovich, Danielle Scherer, Bronna Shapiro Creative Design and Layout: Liora Blum, Liora Blum Graphic Design Creative Direction: John Bishop Special Thanks: Larry Bandemer, Ali Ghazanfari, Helen Greenstein, Matty Kaffeman, Kathy Klimpel, Carolyn Lindsey, Ryan Ragozzine, Matthew Selheimer, Darla West To learn how BMC Software BSM solutions can help your business, call 800-841-2031. BMC Software, the BMC Software logos, and all other BMC Software product or service names are registered trademarks or trademarks of BMC Software, Inc. All other registered trademarks or trademarks belong to their respective companies. Copyright © 2011 BMC Software, Inc. All rights reserved. Focus on: Mainframe Cost Optimization 6 Technology Economics: The Internal Combustion Mainframe Technology Economics reveals that com putational platform choices have economic consequences that must be assessed from a business — rather than a technology — perspective. With such knowledge, it is possible to engineer the “IT cost of goods” for maximum business value. This article shows how. By Dr. Howard A. Rubin 16 The Keys to Effective Mainframe Integration: Automation, Collaboration, and Education Most IT organizations with mainframes have technologies and best practices that can help integrate the mainframe effectively into today’s multiplatform enterprises. Volvo Information Technology found three practices — automation, collaboration, and education — invaluable in overcoming integration challenges. By Per Johansson 22 Four Guidelines for Maximizing the Value of Your Mainframe Wouldn’t it be great if there were a simple formula for maximizing your mainframe and integrating it into a hybrid data center? Of course, it is clear that there is no such formula — because each IT organization is unique in many ways, especially in the nature of the workloads it supports. While this formula does not exist, this article provides four guidelines to help you make the most of your mainframe investment. By Tim Grieser 28 The Mainframe Strikes Back: Five Reasons to Keep Your Applications on the Mainframe Many IT organizations are considering moving workloads from mainframe to distributed platforms to reduce costs. But take a look at the hidden costs of doing so … By Dr. John Shedletsky table of contents 4 A Message from John McKenny Vice President, Mainframe Service Manage ment, BMC Software 34 The Lasting Power of the Mainframe With the growth of mobile and cloud tech nologies, the volumes of data and business transactions that must be processed will only increase. The mainframe will be a critical player in meeting the IT challenges presented by this inevitable growth. Read this article to find out why. By Al deMoya and Nick Pachnos 1 table of contents 40 Keeping Up with Demand While Mainframe Teams Is your IT organization being asked to con stantly provide more and better business services through your hybrid computing environment? You can deliver the business services your company needs while over coming the sometimes daunting problem of complexity. Here are five ways you can stay ahead of demand. By Lilian Vieira The mainframe has a bright future, but only if you continue to leverage your mainframe investment and explain how it maximizes the entire corporate IT invest ment. To accomplish this, practice the seven habits of highly effective mainframe teams — those teams with a track record of successfully leveraging and promoting their mainframes. By Bill Moran 48 Close to Real-Time Delivery of Millions of Messages Thomson Reuters BETA Systems effectively processes 120 to 140 million businesscritical messages each month. This article explains how. By Felix Garcia 54 Flexible, Yet Reliable: Meeting the Challenges of Global Business The needs of global IT users are growing in complexity and diversity. To meet these requirements, use a hybrid solution con sisting of distributed systems integrated with a centralized mainframe. Distributed systems offer flexibility in meeting local user requirements, while the mainframe offers superior reliability and the benefits of centralized data management. By Bruce Chen 2 62 The Seven Habits of Highly Effective Overcoming Complexity 68 Ten Tips for Mainframe Success IT organizations can reduce and optimize mainframe costs by taking advantage of specialty processors, selecting the right vendors, and continually tuning perfor mance. Follow these ten tips to make the most of your mainframe investment. By Jonathan Adams Mainframe Cost Optimization: Delivering the highest levels of service using the least amount of resources. 3 A MESSAGE FROM JOHN MCKENNY Vice President, Mainframe Service Management, BMC Software The mainframe’s unmatched reliability, security, and efficiency in processing large transaction volumes make it critical to many IT organi zations today. But while the mainframe is a key element of many IT infrastructures, 65 percent of the respondents to BMC Software’s 2010 Annual Mainframe Survey indicated that reducing IT costs is a top priority. As a result, this edition of VIEWPOINT, which includes contributions from industry experts, BMC Software customers and partners, and BMC thought leaders, provides advice and practical guidance to help you effectively maximize your mainframe investment and increase the mainframe’s value to the business. In our lead article, Dr. Howard A. Rubin intro duces the concept of Technology Economics to provide a way for IT organizations to make more informed decisions about platform choices, as well as relate compute costs to revenue. Next, Per Johansson of Volvo Information Technology discusses how auto mation, collaboration, and education are the keys to effectively integrating the mainframe into your business operations. 4 The article by Tim Grieser of IDC describes four broad strategies for maximizing your mainframe investment, with the ultimate goal of delivering an excellent end-user experience — end-to-end performance, avail ability, and security — at a reasonable cost. Did you ever wonder if there are hidden costs associated with moving workloads from a mainframe to a distributed systems environ ment? If so, be sure to read the article by Dr. John Shedletsky of IBM, which talks about five hidden costs of doing so. Two articles by BMC partners provide guid ance based on experience with a variety of mainframe and hybrid computing environ ments. Lilian Vieira of CSC BRASIL outlines tips for keeping up with the constant demand for more and better business services while meeting the challenges of increasingly com plex hybrid computing environments. Bruce Chen of Galaxy describes how to best meet the needs of global end users by implementing a hybrid environment that offers flexibility on the front end and stability and centralized data processing with the mainframe. How can you effectively process more than 120 million business-critical messages each month? To find out, read the article by Felix Garcia of Thomson Reuters BETA Systems. And don’t miss the article by Bill Moran of Ptak, Noel, which discusses the characteris tics of teams that have successfully leveraged their mainframe investments. These are just some highlights of the infor mative articles you will find in this edition of VIEWPOINT. We hope the information provided in this publication will help you make the most of your mainframe investment. About John McKenny John McKenny, BMC Software vice presi dent of worldwide marketing for Mainframe Service Management (MSM), leads global teams responsible for the product strategy, direction, and marketing activities for main frame management solutions. He joined BMC in 1995 as a member of the DB2 research and development team, and served as a director in research and development and as senior director of marketing and product manage ment. Prior to joining BMC, he spent 15 years in various IT roles. Best regards, John McKenny Vice President, Mainframe Service Management BMC Software 5 Technology Economics: The Internal Combustion Mainframe Technology Economics: The Internal Combustion Mainframe Technology Economics reveals that computational platform choices have economic consequences that must be assessed from a business — rather than a technology — perspective. With such knowledge, it is possible to engineer the “IT cost of goods” for maximum business value. This article shows how. By Dr. Howard A. Rubin 6 7 Technology Economics: The Internal Combustion Mainframe F or a few moments, don’t even think about information technology. Think about the automobile. There are 750 million to 900 million auto mobiles on our planet: diesel, electric, hybrid, and Hybrid Synergy Drive. And, most of all, automobiles powered by the gasolineburning internal combustion engine. Internal combustion, in use for a century, still powers 99 percent of automobiles. It is predicted to remain the main engine of transportation, with at least 50 percent penetration, through at least 2025. Consider how the internal combustion engine has evolved and has been tuned and refined based on a century of experience. Neither mainframe computing nor server-based computing is right for all forms of computation. This evolution became more obvious to me recently, as I attended the U.S. introduction of the Ferrari 458 in Greenwich, Connecticut. There, under glass, as an integral part of this automotive masterpiece of art and design, was a 275-cubic-inch internal combustion engine producing 580 horsepower, or 2.11 horsepower per cubic inch. It definitely was not my “father’s Oldsmobile” (he actually had a 1970 Buick and, yes, Oldsmobile is gone). His Buick had 350 cubic inches and produced 125 horsepower, or 0.24 horsepower per cubic inch – and probably delivered fewer miles per gallon than the Ferrari. 8 That’s an almost 10x improvement in efficiency over 40 years. It’s a major indication that the industry will not soon discount, discredit, or abandon the internal combustion engine. The Economics of Computational Platforms The same thinking applies in computing. The industry will not soon discount, discredit, or abandon its 50-year history of computing by replacing all computational engines with the “flavor of the day” – servers, blades, or the cloud. But this article is not about mainframe superiority. It is about the economics of com putational platforms and the import ance of Technology Economics in maximizing the value of IT to your business. Most businesses today rely on a mix of com putational platforms – mainframe computers; UNIX, Wintel, and Linux servers; and mid range computers (e.g., AS/400). This mix is mostly un-engineered. Neither mainframe computing nor serverbased computing is right for all forms of computation (and I haven’t even mentioned supercomputing or megaFLOP environ ments). It is essential for organizations to consider both the functional characteristics of their computing needs and the economics. Determining Compute Cost Per $1 Million Revenue The economics include relating compute cost to revenue. Recently, Rubin Worldwide analyzed data from 21 sectors (including gov ernments) and 133 companies. The average company has about 0.37 MIPS and 0.17 servers per $1 million of revenue. In other words, the average $10 billion company has a core platform of 3,700 MIPS and 1,700 physical servers. (As virtualization takes hold, this will change.) Sectors vary widely in their computational needs. At the high end, banking and finance require 10,700 MIPS and 4,600 servers, on average, to support $10 billion in revenue. At the low end, professional ser vices companies require only 1,400 MIPS and 800 servers. If you combine these “platform demographics” with typical values for the average cost of MIPS and the average cost of physical server operation per year, you can determine com pute cost per $1 million revenue (see Figure 1). Sector averages n = 133 companies Industry MIPS per $1 million revenue Servers per $1 million revenue Compute cost (MIPS + server cost) per $1 million revenue Banking 0.98 0.39 $8,505 Consumer products 0.19 0.16 $2,535 Education 0.13 0.05 $1,110 Electronics 0.25 0.11 $2,280 Financial services 1.07 0.46 $9,645 Food & beverage processing 0.18 0.12 $2,070 Government – federal 0.49 0.12 $3,413 Government – state & local 0.38 0.09 $2,655 Health care 0.19 0.13 $2,220 Insurance 0.33 0.16 $3,165 Manufacturing 0.21 0.12 $2,205 Metals & natural resources 0.16 0.12 $1,980 Professional services 0.14 0.08 $1,470 Telecommunications 0.85 0.25 $6,450 Transportation 0.23 0.21 $3,240 Utilities 0.16 0.08 $1,560 Cross-industry average 0.37 0.17 $3,406 Figure 1. Compute cost per $1 million revenue 9 Technology Economics: The Internal Combustion Mainframe For example, a financial services company has a compute cost of $9,645 per $1 million on average, as compared with $1,470 for a professional services company. Note that the analysis uses number of MIPS and number of servers, as opposed to a com mon unit of processing power. The reason is that currently there is no common unit of process ing power. Analysis of Platform Choice To get at the essence of the economic trade- offs between mainframe and server computing choices requires more analysis – beginning with a sector-by-sector analysis of companies’ platform choices. For example, in the study database there are numerous financial ser vices organizations with approximately $10 billion in annual revenue, including a main frame-biased financial services organization with 13,102 mainframe MIPS and 2,359 servers and a server-biased financial services organi zation with 7,643 MIPS and 8,846 servers. Using cost-per-MIPS and cost-per-server data indicates that the mainframe-centric organization has an expense of $83.7 million while the server-centric organization in the same business has an expense of $127.3 million in supporting approximately the same amount of revenue. Of course, making platform and architectural choices requires numerous considerations other than raw economics. However, few organizations even take the time to examine the indicated trade-off. And if you scale this analysis up to a $100 billion enterprise, the 10 cost spread is more than $300 million – that is real money. Extending this analysis across all 21 study sectors indicates that the average $10 billion company would use approximately 3,700 MIPS and 1,700 servers at a cost of $34 million (see Figure 2). A mainframe-biased company would use approximately 4,500 MIPS and 850 servers, at a cost of $29.4 million, while a serverbiased company would use approximately 2,700 MIPS and 3,200 servers at a cost of $45.3 million (see Figure 2; see www.rubin worldwide.com for a table of the elements that shows all this data). Making platform and architectural choices requires numerous considerations other than raw economics. Therefore, the hypothetical average company is $4.7 million or 16 percent higher in expense than the mainframe-biased company. And the server-biased company has $16 million or 55 percent more expense than the main frame-biased company. If business leadership were to see all this analysis, they might ask, “What is the busi ness value of this extra expense?” (Sounds like the issues you hear about power choices for automobiles. And, as with the automotive question, the answer lies in the context and not just with the numbers.) 11 9,800 1,900 1,300 2,500 10,700 1,800 4,900 3,800 1,900 3,300 2,100 1,600 1,400 8,500 2,300 1,600 3,713 Banking Consumer products Education Electronics Financial services Food & beverage processing Government – federal Government – state & local Health care Insurance Manufacturing Metals & natural resources Professional services Telecommunications Transportation Utilities Cross-industry average 1,653 800 2,100 2,500 800 1,200 1,200 1,600 1,300 900 1,150 1,200 4,600 1,100 500 1,600 3,900 Servers 4,041 2,571 1,959 $31,650,000.00 $22,050,000.00 $19,800,000.00 4,546 2,327 $22,200,000.00 $34,064,062.50 4,653 $26,550,000.00 1,959 6,000 2,816 2,204 $20,700,000.00 $34,125,000.00 $15,600,000.00 13,102 $96,450,000.00 $32,400,000.00 3,061 $22,800,000.00 10,408 1,592 $11,100,000.00 $64,500,000.00 2,327 $25,350,000.00 1,714 12,000 $85,050,000.00 $14,700,000.00 MIPS Figure 2. Analysis of platform size and costs by industry sector MIPS Industry Estimated platform cost Platform size and costs to support $10 billion revenue 848 410 1,077 1,282 410 615 615 821 667 462 590 615 2,359 564 256 821 2,000 Servers $29,385,075 $13,124,019 $23,981,162 $60,298,273 $12,021,978 $15,277,865 $18,032,967 $26,799,058 $17,469,388 $25,784,929 $33,192,308 $16,379,906 $83,728,414 $19,698,587 $9,855,573 $19,084,772 $75,000,000 Estimated platform cost -13.8% $(4,705,988) $(2,475,981) $(8,418,838) $(4,201,727) $(2,678,022) $(4,522,135) $(4,017,033) $(4,850,942) $(4,730,612) $(765,071) $(932,692) $(4,320,094) $(12,721,586) $(3,101,413) $(1,244,427) $(6,265,228) $(10,050,000) Gap to “average model” 962 2,115 8,846 2,308 2,212 1,731 2,500 3,077 2,308 2,308 1,538 4,808 4,038 1,538 3,179 7,643 1,286 3,500 2,714 1,357 2,357 1,500 1,143 1,000 6,071 1,643 1,143 2,652 3,077 1,357 929 7,500 7,000 1,786 Servers MIPS $45,313,444 $21,296,703 $49,796,703 33.0% $11,249,382 $5,696,703 $17,396,703 $5,953,846 $13,302,198 $77,802,198 $9,573,626 $29,373,626 $20,653,846 $8,930,769 $30,980,769 $11,264,835 $10,157,143 $32,357,143 $42,914,835 $3,837,363 $4,846,154 $9,316,484 $30,827,473 $30,387,363 $38,971,154 $30,016,484 $127,277,473 $7,447,253 $3,174,725 $14,274,725 $30,247,253 $13,064,835 $25,200,000 Gap to “average model” $38,414,835 $110,250,000 Estimated platform cost Server-centric model Platform size and costs to support $10 billion revenue Mainframe-centric model Platform size and costs to support $10 billion revenue Technology Economics: The Internal Combustion Mainframe The IT Cost of Goods You can bring this analysis down one level to the IT cost of goods (see Figure 3). For example, an optimized mainframe-server mix would reduce the IT cost per stock trade by $.013 – multiply that by a few billion trades and you c a n re a l l y s e e t h e b u s i n e s s v a l u e of optimization. But you can’t make decisions on the econo mics of the platform. Cost per processing second is different on a mainframe, or a server, or even in the cloud, but that’s not a sufficient measurement. You need to measure the business value; for example, the business value of a mainframe’s ability to pump large transaction volumes through for DB2. Industry Measure Airlines Per passenger mile Automotive To make good technology decisions, you have to take a business perspective. If you’re looking at platforms, you have to look at the primary issuer’s cost of goods, time to market, and a couple of other mea surements. Cost of goods is a major one for a transaction-based business. The Future of Technology Economics Technology Economics is clearly still in its infancy. However, with its 50-plus years of history, the mainframe itself is as mature and reliable as the internal combustion engine, and still evolving. Today’s newer platforms will also evolve and will in turn experience competitive pressure as even newer technologies enter the marketplace. Average IT cost of goods Mainframe-biased Server-biased $0.0070 $0.0061 $0.0076 Per vehicle $333.0000 $275.0000 $370.0000 Chemicals Per patent $57,717.0000 $55,800.0000 $59,552.0000 Consulting Per consultant $53,060.0000 $48,900.0000 $62,344.0000 Hospitals Per bed per day $64.3000 $54.4000 $71.7000 Railroads Per ton mile $0.0014 $0.0012 $0.0018 Retail Per store (door) $494,818.0000 $421,346.0000 $560,300.0000 Web sites Per search $0.0420 $0.0460 $0.0041 Trucking Per road mile $0.1770 $0.1550 $0.1940 Armed service Per person $8,036.0000 $6,871.0000 $9,839.0000 Utilities Per megawatt-hour $2.6300 $2.2100 $2.9400 Oil and gas Per barrel of oil $2.1000 $1.7800 $2.3200 Figure 3. The IT cost of goods 12 Decisions about computing need to be made in the context of business relevance. Relevance implies applicability to the busi ness application and a true understanding of econ omics in terms of expense and real value. It is time to explicitly engineer the economics of core computational platforms with the little that is known – that the marketplace doesn’t reward firms that are using only the latest technology at any expense. In the long run, it rewards those that make the optimum use of the right computing res ources in the right way, as evidenced by business performance. clouds, to in-house servers, to traditional mainframes. You’ll see the evolution of bro kering systems that can do provisioning across different environments on demand. You may see the birth of large-scale tech nology commons, with massive sharable resources at known service levels, at defined costs, with dynamic provisioning, so com panies will have the option of doing most strategic work internally. The zEnterprise will bring mainframe and some ser ver platforms under common managem ent. Eventually algorithms may dynamically make the best choice for optimal economics during operation. Many companies today have a server utiliza tion of 6 to 12 percent on a good day. With good common systems management and other kinds of brokering software, and by being able to move things around more fluidly, companies can start to get utilization levels that are consistent and balanced across the platforms. And maybe someday some dynamic brokering – moving the right things to the right place at the right time. In the technology economy, a free market is developing. There are going to be options for doing things – from public clouds, to private Eventually, you will have a tool like a power grid that gets you the best price and processing option – nanosecond by nanosecond – 5 basics of Technology Economics »» Consider both the functional characteristics of your computing needs and the economics. »» Relate compute cost to revenue. »» Analyze the economic trade-offs between mainframe and server computing choices. »» Measure the business value of a computing function. »» Take a business perspective when making technology decisions. 13 Technology Economics: The Internal Combustion Mainframe and brokers it to you. This could be the cornerstone of a core system that will change the economics of all computing. Self-optimizing economics. the business.” Speak in terms of real busi ness-facing classifications: grow revenue, protect revenue, avoid cost, reduce cost, and manage risk. The Time to Start Is Now Then “bite the bullet” and begin to compute return on IT. This is another way of ensuring that IT is viewed in meaningful business terms – as a creator of business value, not a cost. How should CIOs and senior IT profession als begin taking advantage of Technology Economics? The first action is to really “connect the dots” and implement bidirectional transparency. IT infrastructure and development manage ment organizations need to implement the cost accounting to tie what they are doing to platforms, applications, and business products/volumes from the bottom up; then from the top down. The third action is to set up a program to continuously understand the implications of your choices and optimize your choices. Find a way to self-impose continuous discipline. This is not a static market; the workloads are changing and the customer expectations are changing. Historically, the top-down mainframe num bers have looked very high, because the mainframe has been a cost-pool dumping ground. The ser vers ate the meal, but the mainframe paid the check. The space, depreciation, and software suites were allocated to the mainframe. Some orga nizations ar bitrar ily a s signed a high percentage of total data center costs to the mainframe. IT infrastructure and development management organizations need to implement the cost accounting to tie what they are doing to platforms, applications, and business products/volumes. Traditional top-down cost accounting by allo cation naturally leads to false conclusions. A better way is activity-based costing – using granular data from the bottom up to connect the dots. If you start doing it today, you will have an extreme strategic advantage in the short and long term. You’ll make the right platform choices. You’ll make the right moves. You’ll know when to go to cloud. You’ll be dealing in the free market tech nology economy, with tools that no other company has. And that’s a definite com petitive advantage, all around. ● The second action is to change your language and outlook. Implement an expanded port folio model. No longer speak in general terms such as “run the business” or “grow 14 A Huge Opportunity The companies that can master their own Technology Economics, and can master it quickly, will be the real winners. About the Author Dr. Howard A. Rubin is president and CEO of Rubin Worldwide. He is inter nationally recognized for his work as an author, researcher, speaker, and consultant in the areas of IT measurement, techno-business strategy, global software economics, the workforce of the future, the business value of technology, performance measurement and benchmarking, and green IT. Dr. Rubin is a professor emeritus of computer science at Hunter College of the City University of New York, an MIT CISR research affiliate, a Gartner senior advisor, a former board member and executive vice president of META Group, Inc., and a former Nolan Norton research fellow focusing on the development of the balanced business scorecard. In addition, Dr. Rubin has served as a member of the Global Information Economy Working Group of the U.S. State Department’s Advisory Committee on International Economic Policy. He has also worked directly with heads of state around the world to develop national competitive technology strategies for Canada, India, the Philippines, South Africa, and the United States. Dr. Rubin possesses a Ph.D. from the City University of New York in computer science and oceanography. Rubin Worldwide is currently analyzing the microeconomics and macroeconomics of Technology Economics. The firm is building the first “periodic table of the elements,” industry by industry, for IT cost of goods and the linkages between technology costs, total cost of goods, business performance, and even national competitiveness. Visit Rubin Worldwide at www.rubinworldwide.com/rubinww.php. 15 The Keys to Effective Mainframe Integration: The Keys to Effective Mainframe Integration: Automation, Collaboration, and Education Automation, Collaboration, and Education 16 Most IT organizations with mainframes have technologies and best practices that can help integrate the mainframe effectively into today’s multiplatform enterprises. Volvo Information Tech nology found three practices — automation, collaboration, and education — invaluable in overcoming integration challenges. By Per Johansson 17 The Keys to Effective Mainframe Integration: Automation, Collaboration, and Education At Volvo Information Technology, effectively integrating the main frame into business operations is vital to maximizing revenue. Everything the company does is with the goal of providing better service to its customers in the most efficient way. The mainframe’s traditional strengths — reliabilit y, securit y, per formance, and scalability — make it a trusted platform for managing business-critical data, applications, and transactions. More recent developments, such as the new zEnterprise platform, make it possible to stretch the capacity and perfor mance of the mainframe while lowering management and labor costs. Volvo Information Technology is a separate company within the Volvo group of compa nies. Volvo Information Technology provides global IT services to the Volvo group, as well as to other businesses that purchase capac ity and expertise on an outsourced basis. About 40 percent of our business comes from external customers, which span a broad range of industries. Customers value this broad experience and the security of knowing they have the capacity they need, when and where they need it. The mainframe’s traditional strengths – reliability, security, performance, and scalability – make it a trusted platform for managing business-critical data, applications, and transactions. Today, most customers are in the Nordic region, but our business is starting to expand 18 globally. In addition to expanding the busi ness, each new customer enriches the IT team’s experience in managing different kinds of workloads. Integrating the Mainframe into a Multiplatform World Historically, the mainframe had been the dominant server platform at Volvo Information Technology. However, even as many business applications have migrated to UNIX, Windows, and other distributed platforms, the mainframe remains a key component in customers’ busi ness operations. Most notably, the mainframe houses databases. Today, these databases must connect to a growing — and increasingly diverse — set of business applications. In integrating the mainframe into this wider, multiplatform world, the company has faced three challenges: (1) problem detection and resolution is more difficult, (2) a natural cultural gap exists between mainframe IT operations teams and distributed IT operations teams, and (3) getting executive buy-in for the mainf rame is sometimes difficult because of the perception that it is more expensive. In meeting these challenges, three factors — automation, collaboration, and education — have been essential vehicles for success. Simplifying Problem Detection and Resolution Through Automation Meeting customers’ service level agree ments (SLAs) depends on being able to quickly diagnose a problem in a business transaction and find the source — whether it’s in a mainframe subsystem or an IMS table. In earlier times, when workloads ran exclusively on the mainframe, it was usually relatively easy to detect and solve a problem. Today, with pieces of applications residing on different platforms and tapping into the mainframe for data, incident and problem management becomes much more dif ficult. Measuring end-to-end response time for applications is trickier, and identifying the source of a problem can require many different people from inside and outside the mainframe infra structure group. Now, automated tools and processes provide an “aerial view” of business transactions to the service runtime managers, allowing faster problem detection and resolution across platforms. together in an integrated fashion. This means becoming more open-minded to different IT management approaches. Many individuals on distributed teams aren’t aware of the advantages of the mainframe, including its superior reliability, availability, and security. When they find out, they are often interested in how these capabilities can make them more effective. By getting the two teams to collaborate, it’s possible to discover new ways to serve the customers better. Meeting customers’ service level agreements depends on being able to quickly diagnose a problem in a business transaction and find the source. Promoting Collaboration Between the Mainframe and Distributed Systems Teams The infrastructure and operation group within Volvo Information Technology is predominantly organized by platform, but it is moving toward a service run-time management model, where there will be people responsible for whole applications regardless of platform. Service run-time managers work side-by-side with the teams responsible for parts of the infra structure, such as iSeries, UNIX, Windows, storage, networks, and integration platforms such as middleware. New ways of organizing people to better meet customer SLAs pro motes collaboration and a better appreciation of the mainframe’s advantages. As in many IT organizations, the mainframe and distributed teams are separate in Volvo Information Technology. Each team has its own set of tools and practices. However, as the infrastructure platforms become more integrated and shared by different operating systems, the teams must work more closely Eventually, IT staff will be organized into capability layers instead of vertical technol ogy silos. One person will be responsible for hardware, one for operating systems, one for databases, one for transaction han dlers, and so on. There will also be “business Proper automation — automated problem detection and resolution processes that span platforms and teams — is a start to overcoming these challenges. It is essential to continually look for opportunities to raise the level of automation — for example, by building intelligence into processes so that they require little or no intervention by technicians. 19 The Keys to Effective Mainframe Integration: Automation, Collaboration, and Education tracks” — for example, an SAP track — with people responsible for those tracks; each person will decide which resources from each layer to “buy” to meet his or her needs. Dispel Misperceptions Through Education Misperceptions abound about the mainframe. Many think of the mainframe as expensive and unnecessary, particularly when compared to Windows or UNIX servers. To make matters worse, the investment approval process may inadvertently confirm this misperception. The cost of a mainframe investment decision is usually in the millions of dollars; accordingly, senior executives take part in the approval process. In contrast, a distributed systems investment decision is usually in the tens of thousands, but it is one of many such decisions. For example, a company may make 20 separate purchases of 100 x86 boxes at $500 per box. That’s $1 million dollars in 20 separate checks. The net effect is that, on the surface, mainframes may appear to be multimillion-dollar extravagances, and distributed servers may be perceived as lowcost computing power. But in reality, the mainframe is often less expensive than distributed processors. You need to differentiate between total cost of acquisition (TCA) and total cost of ownership (TCO). When investing in a new mainframe, the old one isn’t thrown away — it has tradein value. Behind that big, single number with so many zeros is the hardware, a three- to five-year contract, software licenses, and other value. Plus, the mainframe might be upgraded only every three to four years. It takes fewer people to run the mainframe, and this continues to improve as the man agement tools improve. Spending time presenting facts about the mainframe and its advantages — for example, getting senior executives to consider assign ing building facility costs to each processor group — can overcome misperceptions and objections to further investment. It’s important 5 points to Consider About Mainframe Costs »» Acquisition cost typically includes hardware, a three- to five-year contract, software licenses, and other value. »» The mainframe being replaced has trade-in value. »» A mainframe needs to be upgraded only every three to four years. »» It takes fewer people to manage a mainframe than it would to manage a similar-sized distributed systems environment. »» Be sure to consider both total cost of acquisition (TCA) and total cost of ownership (TCO). 20 to continue to look for better ways to make apples-to-apples comparisons. After these sorts of discussions, executives begin to see the value and will be more open to conversations about how the mainframe can help their businesses. Involving the best technical architects in these discussions will help executives understand exactly how the mainframe will integrate with their operations. In particular, many external customers are deploying new types of applications that can especially benefit from the mainframe. And they want to take advantage of mainframes as their main platforms. Mainframe Integration: Look Beyond the Obvious Mainframe integration is do-able, but it takes some work. Technology — both system-level technology and management software such as cross-platform manage ment tools — continues to improve, which makes the integration job easier. Volvo Information Technology has continued to invest in the mainframe to extend and improve it, adding technologies such as zLinux to make the mainframe efficient and cost-competitive. Currently, 150 Linux serv ers are on the mainframe, covering about 3,500 MIPS. Those servers run some very important applic ations, such as large WebSphere applications. The team is enthusiastic about zEnterprise and the evolution of the mainframe into a hybrid server. This new architecture offers a lot of potential for improving performance by integrating multiplatform applications in a blade rack combined with an enterp rise mainframe. It will offer easier hardware ad ministration, faster internal network speed, and other benefits. A great deal of potential exists for new levels of application integration to businesses like Volvo Information Techno logy, providing new ways to serve customers. But beyond technology improvements, pro cess improvements are equally important. It’s essential to use automation strategically and aggressively to simplify problem detection and management, to deliver better service to customers from current resources. It’s impor tant to promote better collaboration between mainframe and distributed teams, so that they understand the advantages of integrating the mainframe into operations and the value that the mainframe can bring to the distributed side. Finally, it’s important to continually educate senior management about the many advantages of the mainframe — and the value it can bring to the business. ● About the Author Per Johansson is responsible for mainframe infrastructure and operations at Volvo Information Technology. He is a member of BMC’s Mainframe Advisory Council. 21 Four Guidelines for Maximizing the Value of Your Mainframe Four Guidelines for Maximizing the Value of Your Mainframe Wouldn’t it be great if there were a simple formula for maximizing your mainframe and integrating it into a hybrid data center? Of course, it is clear that there is no such formula — because each IT organization is unique in many ways, especially in the nature of the workloads it supports. While this formula does not exist, this article provides four guidelines to help you make the most of your mainframe investment. By Tim Grieser 22 23 Four Guidelines for Maximizing the Value of Your Mainframe M ost enterprises with mainframes are on a journey to continually maximize the value of their main frame investments. This means optimizing mainframe performance to deliver the most business value at the least cost. To deliver the most business value, start by going to the source: the workloads that define and differ entiate your business. How efficiently and cost-effectively you run these workloads — particularly within the context of the hybrid data center — can yield savings and provide competitive advantage. Which Workloads Should Run on Which Platforms? By following four broad guidelines, you can effectively determine, on a case-by-case basis, which workloads to run on which platforms. These guidelines are (1) analyze each workload, (2) consider your staff’s mix of skills, (3) make the best use of tools, and (4) carefully plan capacity. And, while you are implementing these guidelines, always remember that the ulti mate goal of maximizing the mainframe and integrating it into a hybrid data center is to deliver an excellent end-user experience — end-to-end performance, availability, and security — at a reasonable cost. 1. Analyze Each Workload Start by analyzing each workload. This is by far the most important guideline. This guide line is intuitive to most people who see the mainframe as another class of server that exists in a multi-server environment within a data center. And most people understand that certain types of applications and certain 24 types of workloads operate most efficiently and most cost-effectively on certain types of servers. This guideline may sound obvious when it’s spelled out, but is also one step that is often overlooked. The workloads often supported on the main frame include the following: »» The traditional mainframe workloads of high-volume transaction processing against large databases, particularly financial services applications; and batchprocessing workloads. »» The recently popular (during the last five years or so) workloads of Linux-based applications and Java-based applications. The mainframe increasingly plays a role as part of a hybrid application deployment (a hybrid application is an application with one or more tiers on the mainframe and other tiers on distributed platforms). In these deployments, the mainframe is especially effective as a database server. The ultimate goal of maximizing the mainframe and integrating it into a hybrid data center is to deliver an excellent end-user experience at a reasonable cost. These are common uses of mainframes, generally speaking. But when you analyze your workloads, don’t generalize. Be specific; take your analysis down to specific work loads. Answer questions such as: Which workloads is the mainframe supporting and how effectively? What is the number of accesses or users using those workloads? What are the costs of supporting those workloads? And how does that compare to spreading out the work across another platform or platforms? and operational practices — and sometimes deeper topics. Always look at things on a workload basis. In the near future, vendors will increasingly be discussing workload optimization and most likely will provide even further optimization capabilities than are available today. As long as you have a source of deep main frame expertise on staff, this can work reasonably well over the near term. In the longer term (five to ten years), though, as the veteran mainframe professionals retire, you may need to rely on vendors or other outside services, at considerable cost. Or, of course, new tools may be available by then to fill the gap. 2. Consider Your Staff’s Mix of Skills The mix of skills on your staff, currently and in the near future, is an important criterion for platform selection because it partly deter mines the cost of supporting the workload. Some IT organizations are suffering from a lack of a skilled mainframe staff and are facing the prospect of needing to hire main frame experts at high costs. They are looking for ways to control or reduce those costs. One way to control costs is to take advantage of nontraditional sources on the mainframe, e.g., Linux and Java, or use tools that have more modern interfaces, such as Webbased interfaces, and don’t rely so much on knowledge of traditional mainframe greenscreen interfaces. The mix of skills on your staff, currently and in the near future, is an important criterion for platform selection because it partly determines the cost of supporting the workload. Another way to control costs is to train existing and newly hired staffers. In many organizations, the mainframe veterans for mally or informally train new mainframe administrators to become apprentice mainframe specialists. They teach them the basics — technology terms, attributes, 3. Make the Best Use of Tools Always select and use tools with the goal of maximizing the end-user experience. The most obvious class of tools is event manage ment tools. Every large data center needs to be able to make multiple levels of response to problems and outages. The problems and outages will be unique to each data center, depending on the mix of workloads and the mix of platforms. Every large data center needs to be able to make multiple levels of response to problems and outages. The ability to deliver excellent service to an end user also requires the ability to manage across the multiple platforms on which the end user’s application is deployed. The plat forms will often include mainframe tiers and UNIX, Linux, or Windows tiers. You need endto-end transaction management software that successfully supports all the platforms, so that you can monitor and remediate any kind of event that might interrupt or degrade service, regardless of the platform on which it occurs. In other words, you need a holistic 25 Four Guidelines for Maximizing the Value of Your Mainframe approach to application performance and service management. Your event management tools should also support, or even enhance, the mix of skills on your staff. General practitioners can solve many lower-tier problems, but as they trace a problem into a technical area, they must defer to practitioners with the technical expertise to go after and fix that problem. MIPS consumption is expensive in terms of software license costs, and there are ways — such as offloading to specialty processors — to avoid the cost of MIPS and reduce software costs. Tools such as simplified, Web-based inter faces may be able to help your general practitioners go deeper into silos without having too many silo skills. Event manage ment tools are evolving toward allowing general practitioners to solve more kinds of problems than they currently do. 4. Carefully Plan Capacity Capacity planning tools and performance management tools can help you carefully size mainframe capacity, particularly proces sor capacity, to fit your current and near-term workload needs, and help you avoid buying excess hardware capacity. Capacity tends to come in large increments, in terms of costs; therefore you can save substantially if you can reduce the size of the increment you must buy and delay the date at which you must step up and buy it. You can also leverage specialty processors. They reduce general-purpose mainframe MIPS consumption by individual applications and even by utility and management soft ware. MIPS consumption itself is expensive in terms of software license costs, and there are ways — such as offloading to specialty processors — to avoid the cost of MIPS and reduce software costs. 5 questions to Ask When Analyzing Workloads »» Which workloads is the mainframe supporting? »» How effectively is the mainframe supporting those workloads? »» What is the number of accesses or users using those workloads? »» What are the costs of supporting those workloads? »» How do those cost s compare to sprea ding out the wor k acros s another platform or platforms? 26 It’s All About the End-User Experience Most data centers will continue to be hybrid for a long time. Bringing the platforms closer together and managing applications as a whole across multiple platforms becomes more and more important to delivering and demonstrating value from your mainframe. reasonable cost. Excellent end-user experi ence becomes even more important as you evolve — for example, as you add cloudenablement and a lot more Web access. Your data center must continually become more end-user focused, and must become more capable of measuring and managing the holistic experience. ● Remember the ultimate goal is an excellent end-user experience — end-to-end perfor mance, availability, and security — at a About the Author Tim Grieser is program vice president, Enterprise System Management Software, at IDC. He is responsible for system management software research in IDC’s Enterprise System Management Software program. His coverage includes software for managing systems and applications across a wide variety of platforms. A key focus area is e-business and distributed applica tions performance and availability, especially Web applications response time from the end-user perspective. Grieser has also served in other key research management positions at IDC. Before joining IDC, he had 20 years of experience in the development and marketing of performance and availability management software. He was vice president of North American Marketing at BGS Systems, later acquired by BMC Software. Grieser has graduate degrees in computer science, specializing in computer performance analysis. 27 Five Reasons to Keep Your Applications on the Mainframe The Mainframe Strikes Back: Five Reasons to Keep Your Applications on the Mainframe The Mainframe Strikes Back: 28 Many IT organizations are considering moving workloads from main frame to distributed platforms to reduce costs. But take a look at the hidden costs of doing so … By Dr. John Shedletsky 29 The Mainframe Strikes Back: Five Reasons to Keep Your Applications on the Mainframe W ith a relentless focus on cost reduction, many IT organizations are considering moving appli cations from the mainframe to distributed environments with the expectation of saving money. Mainframes are big. Their purchase and operating costs are often very visible. As a result, they become a frequent target for cost reduction. Individual distributed systems have a lower cost of acquisition, and their total cost of operation is often not readily visible. The conventional thinking is that mainframes are expensive to run, so moving mainframe workloads to distributed servers will cut costs. IBM has conducted extensive on-site analyses of the cost benefits of moving a wide range of critical applications from mainframe to distributed environments. These analyses include the audited actual costs after appli cations were moved. These studies reveal a conclusion that is counter to conventional wisdom: In 48 of the last 50 studies conduc ted, it actually cost less to stay and grow on the mainframe than to move to a distributed systems environment. Workloads running on the mainframe are often a best fit for that platform, and admin istrators achieve significant operational efficiencies on a consolidated mainframe platform. Alternative platforms often cost more. In fact, on average for the 48 cases studied, staying on the mainframe was 42 percent less expensive over a given time period. The bottom line is that the actual cost of moving applications from mainframe to distributed environments may be greater than you think. 30 The mainframe seems more expensive, but in many cases it’s not. What drives the misperception? Five Hidden Costs Five hidden costs often inflate the actual cost of moving mainframe-based applica tions to a distributed environment: core proliferation, code expansion, systems management creep, functional separation, and disaster recovery. 1. Core Proliferation IT organizations sometimes use simplistic benchmarks to compare performance on mainframe and distributed systems. Basic performance tests might suggest that you can replace a single processor on the main frame with a single processor core on a distributed server. However, these bench marks don’t accurately reflect complex application performance. As a result, you might need multiple distributed cores to get the same performance as a single processor running an application in the mainframe environment. One customer found that dozens of distributed cores on multiple servers were needed to replace a mainframe application using only five processors. In 48 of the last 50 studies conducted, it actually cost less to stay and grow on the mainframe than to move to a distributed systems environment. Core proliferation drives up costs because you need to buy more distributed servers. The servers that are appropriate to run heavy workloads are typically high-end, multi-core servers. However, the biggest cost driver is often software, which is typically priced per core in a distributed environment. For an enterprise database or other application that is fairly expensive on a per-core basis, the inflation of software license costs can be quite substantial. So core proliferation in distributed systems results in increased costs for both hardware and software. Mainframe failover capacity is provided at a small fraction of the cost of the primary production capacity. 2. Code Expansion 4. Functional Separation The amount of code required to run a particu lar application in a distributed environment can often expand by several times. As a result, you need more instructions to per form the same amount of work. Increased instruction volume increases the required hardware capacity, which can also inflate software license costs. Many IT organizations have separate produc tion, development, and quality assurance environments — all running in isolated, logical partitions on the mainframe. Workloads in the non-production environments are typically variable and may reach peak load when IT is managing new releases. Each environment shares underlying processing resources, so the additional cost of each resource in the mainframe environment is small as a per centage of overall resource usage. But in a distributed environment, all three often get their own set of distributed servers. The addi tional cost of redundant environments is high. Functional separation adds to the proliferation of cores and software license inflation. Inefficient compiler technology contributes to code expansion. Recompiling COBOL applications to distributed platforms has been observed to produce up to eight-times code expansion. Rewriting COBOL applica tions on CICS to Java applications on a Java transaction engine can typically produce a three-times code expansion. tools are often priced by core, you can actually end up spending more for systems manage ment software in the distributed environment. 5. Disaster Recovery Moving from a single logical database image on DB2 for z/OS to a clustered solution on Oracle RAC may require more cycles. When moving from an IMS hierarchical database model on the mainframe to a relational database model in a distributed environment, three-times code expansion is also common. 3. Systems Management Creep To replicate mainframe systems management capabilities in the distributed environment, you may need licenses for a wide range of systems management tools. Since these For critical workloads that have a high cost of downtime, you probably deploy duplicate failover sites to use in the case of a serious outage. Typical mainframe pricing includes very deep discounts for the failover site servers and a nominal recurring fee for the right to exercise the backup processors for testing purposes. The result is that main frame failover capacity is provided at a small fraction of the cost of the primary production capacity. This is advantageous when com pared to the price of building and maintaining a failover site using distributed systems, 31 The Mainframe Strikes Back: Five Reasons to Keep Your Applications on the Mainframe which often do not offer hardware and soft ware discounts for the failover environment. “Best Fit” Delivers Optimal Price/Performance However, there are many other types of work loads that data centers may need to support. When considering price/performance, it may make sense to run these workloads on Intel-processor-based or UNIX environments. Recognizing this fact, IBM has recently announced the zEnterprise System. This sys tem is actually a hybrid, offering four different execution environments in a single system of systems. You can choose to run workloads in z/OS, Linux on z/VM, or AIX on Power blade environments. A statement of direction also includes Linux on Intel blades. Most of the workloads currently running on z/OS are well suited for the mainframe computing environment. They might have high transaction scale; a large, single database image; or significant I/O band width requirements; or they might run intensive batch processing jobs. For the reasons discussed previously, moving these types of workloads ends up costing more and may reduce performance or quality of service. So from a price/perform ance standpoint, it makes sense to keep these workloads on the mainframe. You can evaluate each specific workload to determ ine which environment in the zEnterprise platform will provide the opti mal price/performance. Most existing z/OS workloads are already achieving optimum price/performance on z/OS. A light process ing workload may be best suited for the Intel blade. A Power blade may be the right choice for a heavy processing workload. A workload with light CPU and heavy I/O may be a best fit for the Linux environment on z/VM. Most workloads can run on multiple environments, You may optimize by using the distributed failover capacity for development and quality assurance workloads. In the case of an actual site failover, these workloads would be deferred while the production workloads were run. But the net effect is that you still pay twice the primary production cost. 5 Hidden Costs of Moving to a Distributed Environment »» Core proliferation »» Code expansion »» Systems management creep »» Functional separation »» Disaster recovery 32 but the best fit is the environment that satis fies the quality of service requirements and minimizes price/performance. efficiency of systems management tasks. zEnterprise extends the ability to apply these practices to a broader set of workloads. Reduce Management Costs Combine the Best of Both In the recent past, the productivity of main frame management staff has increased dramatically. IBM has kept track of this trend during cost consulting studies with many customers. The data collected suggests that on average, the number of MIPS per full-time equivalent (FTE) has increased 63 percent in the last seven years. During the same period, the average improvement in the number of servers per FTE in a distributed environment has been only 16 percent. The ideal solution is to combine the benefits of a “best fit” workload portfolio with the efficiencies of a consolidated platform. zEnterprise enables data centers to do both. Instead of pursuing a strategy to move appli cations off the mainframe, organizations should move applications to the best-fit environment on the zEnterprise platform and use the common management framework to support structured management practices for all the workloads. The best-fit strategy optimizes price/performance, while structured management practices reduce labor costs. The culture of structured management is the biggest reason for the mainframe productivity advantage. Consolidating many workloads onto a single platform, with common manage ment tools, is known to produce significant labor productivity gains. Early mainframe support for virtualization and consolidation led to the practice of standardizing software stacks for virtual machines. Quality assur ance testing is a mainframe practice that reduces the possibility of releasing problem code to production. Automation of repetitious tasks can also reduce labor requirements. The integrated nature of the mainframe plat form enables these practices to increase the Stay with the Mainframe Don’t rush to move workloads from the main frame to a distributed environment. Analyses show that a combination of a “best fit” hard ware platform and structured management practices gives zEnterprise the best price/ performance when considering a portfolio of workloads. So, if you are considering moving workloads from the mainframe to a distrib uted environment, be sure also to evaluate the benefits of consolidating more distributed workloads on the zEnterprise platform. You might be surprised by the results. ● About the Author Dr. John Shedletsky is vice president of competitive technology at IBM. He leads an organization of systems experts whose primary mission is to identify and quantify the benefits of IBM workload optimized systems. 33 The Lasting Power of the Mainframe The Lasting Power of the Mainframe With the growth of mobile and cloud technologies, the volumes of data and business transactions that must be processed will only increase. The mainframe will be a critical player in meeting the IT challenges presented by this inevitable growth. Read on to find out why. By Al deMoya and Nick Pachnos 34 35 The Lasting Power of the Mainframe F lash back to 25 years ago in the banking industry. At that time, a state-of-theart business service was consolidating customer data from multiple bank accounts (savings, checking, trust, and so on) into sin gle, unified statements that were mailed to customers once a month. Today, you can take a photo of the front and back of your check with your smartphone, submit it to your bank electronically, and the check is automatically deposited into your account. Now that’s really state of the art. Soon you will be able to send a text message to a vending machine and have your selection pop out. With this technology comes the need to ensure that every single text message is captured and processed. A wireless service provider may deliver billions of text messages daily. Sure, many of those text messages will be from teenagers saying “hey” back and forth. But, now, text messages may also include financial transactions. As a result, the wireless provider will need to capture the vending machine purchase and add the cost to the purchaser’s bill — otherwise, the wireless provider will be responsible for paying the charge. This capability will amplify the data volume that needs to be processed, as well as expand the revenue-generating capability of wireless providers. Some major insurance companies now offer applications for mobile devices that allow you to read your policy online and provide you with an image of proof of insurance. If you are involved in a car accident — not your fault, of course — an application provides guidance on what to do. You may even be able to collect the other driver’s insurance information, take 36 photos, and immediately submit this informa tion with a claim via your mobile device. Data captured within minutes of the accident is instantly received by the insurance company, speeding up the claims process. The mainframe is still the best data processor, and it’s the optimal platform to handle the combination of very high data and transaction volumes we see today. The common theme for all of these scenarios is that they involve very large volumes of data and a huge number of business transactions that must be processed by a reliable, secure platform. This is why the commercial trans action workhorse for the largest financial institutions, insurance companies, phone companies, and so on is the IBM zSeries mainframe platform. More than 50 years after its creation, the zSeries mainframe is still a key element of the data center, as it continues to keep up with new technologies and new business requirements. It’s All About the Data The way we conduct business has changed over the years, but the fundamentals remain the same — namely, the need to quickly, economically, reliably, and securely process and capture each and every business trans action. In addition, companies need to reliably update customer data, vendor inventory, and other essential records so that they can sup ply goods and services, bill customers, and be paid. Regardless of where the transaction originates, the mainframe is at the back end, keeping up with the enormous volume of transactions and then accurately posting those transactions to the huge volume of data it holds. Quite simply, the mainframe is still the best data processor, and it’s the optimal platform to handle the combination of very high data and transaction volumes we see today. Why Does the Mainframe Continue to Be Critical to the Business? It’s clear that the mainframe is key to many businesses. Some of the largest corporations in the world may need to process billions of transactions a day. Consider a company that processes 1 billion transactions a day. That’s over 40 million transactions per hour, or 11,000 transactions per second. The main frame is capable of processing this level of transactions while enabling the data to be accessed 24x7x365. And yet, despite its reliability and security, the mainframe seems to be constantly defending its position. So, what are the benefits of the mainframe? Here are just a few. 1. Centralized Data: One Master Customer Record Most companies keep their most critical production data in one place. Many corpora tions have grown very large through industry consolidations and, consequently, the amount of data that must be processed is incredibly large. Add the fact that many industries have new compliance require ments that necessitate retention of customer data for longer periods of time. These factors contribute to the growth in data volume. Now consider that customers expect in stantaneous access to their information (accounts, transactions, and so on) 24 hours a day, 7 days a week, 52 weeks a year. For example, the package-shipping business needs to track your package shipment from the moment it is picked up through multiple consolidation and distribution centers to its final delivery to your doorstep. Then they capture your signature to close the record. Most industries have experienced similar explosive growth in transaction volume. The mainframe’s centralized environment is the most reliable, secure, and effective plat form for handling large amounts of data and ver y high transaction volumes. As the Internet opens up businesses to more and more people who expect immediate access 24x7, it is essential to store data on a platform that offers unmatched performance and availability, i.e., the mainframe. The mainframe offers a higher level of continuous availability than other platforms. 2. High Availability The ability to handle large data volumes and transactions is a continuous requirement. The platform must be available 24 hours a day because business transactions do not sleep. The mainframe has matured to the point that most maintenance processes can be performed while the data is online and available for update. As a result, the main frame offers a higher level of continuous availability than other platforms. Many companies have gone multiple years without experiencing a maintenance outage on their mainframe systems. 37 The Lasting Power of the Mainframe 3. Reliability The mainframe platform is exceedingly reli able because of the combination of redundant mature hardware, a robust operating system, and extremely reliable database software. The mainframe can process transactions at significantly higher transaction volumes than other platforms. You can even drive CPU utilization to peak levels without causing a significant slowdown. In regard to efficiency, reliability, and service ability, the mainframe was designed from the outset to handle multiple commercial workloads. Over the decades, engineers have constantly refined the firmware that allows internal computers to serve the data to the actual processing unit. This continual refinement on the same basic architecture is a major differentiator for the mainframe. 4. Adaptability of Middleware Layer Previously, the process of getting transactions into the mainframe was fairly rigid, as a result of the mainframe’s closed, self-contained architecture. But the Internet changed the mainframe world. Organizations started moving heavily to client/server and Web-based interfaces. And it seemed, for a time, that the mainframe was doomed. But a new layer of middleware software was developed to enable the mainframe to connect to the new Internet technologies. It’s this middleware layer that allows the main frame to interface with all kinds of devices — smartphone, personal computer, or any device that supports an Internet browser. Now, with middleware infrastructure, the mainframe can easily handle some very innovative data types and transactions. This middleware infrastructure has opened the door to trans actions that in the past were quite restricted. 5. Cost Efficiency For companies that have already made a main frame acquisition investment, research shows that the mainframe is the more cost-efficient platform in IT. The mainframe’s architecture enables it to be highly scalable, without the need to add more staff. And the mainframe offers security, availability, and recoverability that cannot be matched in a distributed systems 5 reasons to Keep the Mainframe as the Mainstay »» Enterprises have very large volumes of data — and those volumes are increasing. »» Enterprises have an enormous number of business transactions. »» Data and transactions must be processed by a reliable, secure platform. »» The mainframe is still the best data processor. »» The mainframe continues to keep up with new technologies and new business requirements. 38 environment. Without significant investment in redundant hardware in a distributed environ ment, you can’t get close to the recoverability that you have with the mainframe. The mainframe offers security, availability, and recoverability that cannot be matched in a distributed systems environment. Additionally, the distributed environment consumes a significant amount of power, cooling, and floor space. So, while you can heavily invest in a distributed environment to attempt to achieve the same level of avail ability that you would have on the mainframe, you will invest a hefty premium and your costs will skyrocket. An often-overlooked advantage is that the mainframe requires fewer support staff than a matching distributed environment. Main frame growth does not result in a matching growth in staff, as is often the case with other IT processing platforms. Moreover, the scal ability of the mainframe can allow you to grow your business with a relatively small incre mental increase in costs. 6. Low Risk For companies that already have a main frame, moving business-critical data to another platform can be risky and is a huge project requiring a significant amount of time and staffing resources. When you already own a system that is doing its job, is available 24x7, and is as efficient as the mainframe, why move data to a system that may not be as reliable? The Mainframe Is Here to Stay IT continues to evolve, allowing the end user immediate 24x7 access through the Internet, the cloud, and other advanced technologies. As long as the need exists to process incredibly large transaction volumes with very large databases, the mainframe will continue to be the platform of choice for most companies. How long will the mainframe be around? Quite a long time indeed. ● About the Authors Al deMoya has been in the IT industry for 26 years and has been with BMC Software for the last 21 years. In his current role, deMoya is responsible for assisting the BMC field organization to better understand BMC Software’s portfolio of mainframe database and performance solutions. Previously, deMoya managed marketing, sales, and development teams for BMC and was a systems engineer for IBM, specializing in mainframe hardware and software systems. Nick Pachnos has been in the IT industry for 27 years and at BMC for the last 13. In his current role, his responsibilities include managing the technical marketing and sales enablement functions for BMC’s Mainframe Data and Performance Management solutions. He also manages BMC’s Mainframe Executive Advisory Council. Pachnos previously managed BMC’s mainframe professional services practice and was a regional sales manager. 39 Keeping Up with Demand While Overcoming Complexity 40 Keeping Up with Demand While Overcoming Complexity Is your IT organization being asked to constantly provide more and better business services through your hybrid computing environment? You can deliver the business services your company needs while overcoming the sometimes daunting problem of complexity. Here are five ways you can stay ahead of demand. By Lilian Vieira 41 Keeping Up with Demand While Overcoming Complexity In Brazil, as in most countries, IT organizations are striving to keep up with the constant demand for new and better business services. They are developing and delivering those business services in hybrid computing environments (mainframe platforms plus distributed plat forms). For smooth operation and flexibility, they are developing Java-based applications for use on IBM’s WebSphere Application Server to operate cross-platform. A major challenge is that these hybrid com puting environments are complex and become more so as you add new applica tions. Fortunately, today there are effective ways of meeting this challenge. You can quickly and accurately determine the causes of bottlenecks and failures in the hybrid computing environment and ultimately ensure that your applications always operate as expected. The following five tips can help you meet the challenge. Five Tips for Success in the Hybrid Computing Environment From its recent experience in guiding compa nies to greater IT effectiveness, CSC BRASIL offers the following impor tant tips for companies who want to (1) deliver more and better business ser vices in the hybrid computing environment; (2) manage an environment that provides big business benefits but at the same time brings with it a highly complex set of components that must all work together to deliver value (this environment spans platforms and applications, which means it is much more difficult to quickly pinpoint and resolve problems); and (3) control costs. 1. Value Your Mainframe First, recognize the real value of your main frame. There is a common myth that the mainframe is becoming obsolete and is costly. But in fact, mainframe technology has always delivered high reliability, avail ability, security, and flexibility. The mainframe now delivers those same capabilities but in even greater abundance. Today’s mainframe is the platform that hosts the most powerful tools for monitoring, managing, and automating. Today’s mainframe is the platform that hosts the most powerful tools for monitoring, managing, and automating. These tools simplify the management of business appli cations and help deliver reliable business services. And mainframe price/performance continues to improve. For example, IBM : »» Provides a major advance in costeffectiveness by offering specialty engines that can offload DB2 processing and other workloads from expensive, general-purpose central processors, and specialty engines that optimally execute Java applications (CSC BRASIL’s customers are currently acquiring more of these engines) »» Offers the new zEnterprise System that helps reduce energy consumption by up to 80 percent1 1 “Announcing the IBM zEnterprise System,” IBM, www-07.ibm.com/systems/in/z/news/announcement/20100722_annc.html. 42 »» For Linux workloads, delivers up to 60 percent more performance at 35 percent lower price2 By taking advantage of the mainframe’s continuing evolution, you can streamline service deliver y and constantly achieve better price/performance. 2. Get an Integrated View In a hybrid computing environment, most transactions flow through many systems and subsystems — multiple silos such as z/OS, CICS, DB2, and middleware. Trans actions can slow down or fail in any of these areas, in TCP/IP, or while moving between the mainframe and distributed systems. In this environment, it can be difficult to identify the location and cause of a bottleneck. But don’t be intimidated, as you can tame the complexity if you understand the flow: where something starts, what it passes through, and what systems it must touch until it gets to its endpoint and delivers a response to the user. You can acquire an integrated set of tools to monitor and manage both across and within the flow. Comprehensive monit oring and manage ment solutions give you an integrated view of all transaction flows; they can put all your applications and the components that sup port them on “a single pane of glass.” These tools can also give you the ability to find and fix bottlenecks anywhere in the environment. They can help you enter and leave each silo quickly, without wasting a lot of time keeping track of where you are and without getting lost — and proceed directly to the cause of any problem. 3. Automate as You Go Have an automation platform and an automation strategy. Your monitoring and management solution should have an automation platform. This should include the right event management system to make everyone’s job easier; however, suc cessfully implementing such a solution requires an understanding of the environ ment. Best practices say that you must determine the following: »» Which metrics to collect, out of the large number of available metrics »» How to ensure that systems and applications are running properly »» How to determine the appropriate threshold settings for the metrics »» What steps to take when a threshold is reached »» How to immediately notify the right stakeholders when problems arise »» How and where to automate Have an automation platform and an automation strategy. By following these steps, you will be alerted to potential problems — in effect, you will be creating a comprehensive early warning system. When you reach that point, you will have the right automation platform. You will 2 Ibid. 43 Keeping Up with Demand While Overcoming Complexity be resolving most problems automatically, without getting people involved. solutions? What is the most cost-effective solution, considering all the factors? Be sure to look at the business needs and the desired business results when selecting the platform or platforms. Quite simply, if you make a platform decision without considering business needs, it may be difficult to actually meet all the requirements of the business. Consider the complete business requirements and answer the relevant questions, and then determine the correct infrastructure to meet the overall requirements. However, as you are progressing to that point, and even after you have reached it, you will occasionally need to determine what to do about new kinds of problems that arise. This is where your automation strategy comes in. Your strategy should be as follows: When ever you encounter a new kind of problem, strive to fix the problem rapidly, and fix it, if possible, in a way that will allow the problem to be remediated automatically when it recurs. Don’t just do a manual fix and then hope (often unrealistically) for the quiet period in which you will have ample time to auto mate everything. 4. Consider Business Needs When Selecting Your Platform or Platforms Be sure to look at the business needs and the desired business results when selecting the platform or platforms. In other words, don’t buy a platform and try to retrofit it to meet business needs. You’ll want to fully under stand the overall business requirements and user needs before you select the platform. For example, be sure to answer questions such as: What response time is required? How many users will need to access the business applications at the same time? What level of availability does the business require? Does the platform need to be scalable to accommodate future business needs? How fast do you need to roll out 44 5. Conduct Comprehensive Testing Before Rollout of New Applications Conduct extensive testing before rolling out new applications. And of course, make sure that you have the tools in place to effectively monitor and manage new applications. This due diligence should be done early in the process. The use of analysis and predictive tools as well as monitors will make your testing process even more effective. Remember, whenever a new application goes into production, the whole environment — and the user community — can be affected. You’ll want to ensure that you have run as many potential simulations as possible before implementing a new application. This means that you will want to ensure that your test environment is as robust as possible. That may require an investment in people, software, and hardware. It will also mean that care is taken in the creation of test scripts and/or cases to be sure that they emulate the production environment and load as much as possible. The effort and cost will be well worth it in the long run. With comprehensive testing, you can be more confident that the application can handle the required workload with the necessary response times, and that few, if any, bugs exist. Ensure that your test environment is as robust as possible. You will want to use your performance and monitoring tools not only in production but also as part of the testing environment. The use of analysis and predictive tools as well as monitors will make your testing process even more effective. Use these tool sets to ensure that when the application goes live, you will know that its performance will meet the needs of the business. Keeping Up with Demand: Two Examples How are organizations addressing these issues in the real world? Here are two examples. A Multiplatform Government Agency One of CSC BRASIL’s largest customers is a Brazilian government agency responsible for hosting government applications. The agency’s IT organization faces the challen ging task of providing reliable and secure IT for a growing number of applications while continuing to control costs. Years ago, this agency’s IT organization dev elo ped and ran applications on the mainframe only. More recently, the agency has been developing and running applica tions on distributed platforms as part of the free software initiatives but has continued to host applications on the mainframe. A few years ago, the agency had moved some existing applications from distributed platforms to the mainframe and had also upgraded some select mainframe appli cations to reside on distributed platforms. Today, the agency develops and runs many 5 questions to Ask When Selecting a Platform »» What response time is required? »» How many users will need to access the business applications at the same time? »» What level of availability does the business require? »» Does the platform need to be scalable to accommodate future business needs? How fast do you need to roll out solutions? »» What is the most cost-effective solution, considering all the factors? 45 Keeping Up with Demand While Overcoming Complexity of its applications cross-platform. This approach helps deliver services in a more cost-effective way and delivers the high level of data security that the government demands. The agency runs a great number of Java WebSphere applications both on the main frame and on distributed platforms. These applications access databases and process data on the mainframe, then return to the distributed side. The agency has established an open environment on the mainframe side and the distributed side. The cross-platform approach has allowed the agency to add new applications more rapidly. One major benefit of this openness is that the agency is free to run applications where they will deliver the best quality of service to the government. This is a much more productive and streamlined way of oper ating, as opposed to being restricted to keeping certain applications on the main frame and certain applications on distributed systems, as the agency had to do in the past. The cross-platform approach has allowed the agency to add new applications more rapidly. Two years ago, for example, the total number of Java applications running on z/OS was 68. It is now more than 100. With so much new and powerful technology available, the agency can take advantage of many new ways to meet increased demand while controlling cost and complexity. 46 An Outsourcer Maximizing Capacity Another large customer of CSC BRASIL is an outsourcer that always strives to use its mainframe capacity to offer more types of services to more customers in an effort to realize additional revenue. The outsourcer has been very successful in finding ways to use its IBM System z10 mainframe to its maximum capacity. For example, the company has been offering Linux on the mainframe at an attractive price. The outsourcer’s customers gain the benefits of running Linux within the virtual machine (VM) environment; they can run multiple images — even in the hundreds or thousands — which can reliably share resources and high-speed communications. Now the outsourcer plans to use the new IBM zEnterprise System to deliver cloud services. The company has successfully signed up customers in advance — a private cloud from a reliable outsourcer is a service that seems to appeal to many customers. Today’s mainframes are more powerful, more flexible, more secure, and easier to use than ever. The new system gives the outsourcer an opportunity to offer a wider range of oper ating systems to its customers. The system appears to be very flexible and very accom modating for customers who want even more mainframe security and stability. Delivering Increased Value Today, you have many opportunities to deliver more value to the business. And there are many tools and technologies to help you overcome the complexity that is part of delivering powerful applications. The mainframe can be your ally. Today’s mainframes are more power ful, more efficient, more flexible, more secure, and easier to use than ever. And the mainframe works more smoothly and economically with your distributed systems. To keep up with the demand to deliver more business services, be sure to take an inte grated view of the mainframe and distributed systems. Leverage automated solutions to create an early warning system of potential issues as well as to automatically fix problems that may arise. Be sure to understand the business requirements before selecting a platform. Finally, before you roll out a solution, ensure that it has undergone comprehensive testing. By following these guidelines, you will be well on your way to delivering more — and more effective — business services. ● About the Author Lilian Vieira has a degree in business administration and has several spe cializations, including an M.B.A. from the Fundacao Instituto de Administracao, University of Sao Paulo. She has 28 years of mainframe experience, working as a system consultant for several Brazilian companies. For the past several years, she has worked as MSM Business Support Manager for CSC BRASIL. 47 Close to Real-Time Delivery of Millions of Messages Close to Real-Time Delivery of Millions of Messages How does Thomson Reuters BETA Systems effectively process 120 to 140 million business-critical messages each month? Read on to find out … By Felix Garcia 48 49 Close to Real-Time Delivery of Millions of Messages G etting the right information — and a lot of it — to clients or end users in a timely manner is a big priority for many large enterprises. At Thomson Reuters, the middleware messaging infrastructure enables flexible and reliable messaging that helps our clients and firms make critical, market-leading business decisions. Thomson Reuters combines industry exper tise and technology to deliver information to businesses and professionals in the finan cial, legal, tax and accounting, healthcare, science, and media markets. The company has more than 55,000 employees in more than 100 countries and posted 2009 revenues of US$12.9 billion. To our clients and firms in the financial markets, Thomson Reuters provides content, news, and data-feed solutions that help close deals. The company offers several different types of brokerage messages, including trade executions, cash entries, security movement, security master file updates, a nd completion aler t s. T he c o m p a n y a l s o p rov i d e s mi d d le w a r e processing of vendor-related messages from Nasdaq, the New York Stock Exchange, DTCC Omgeo, and others, processing 120 to 140 million messages each month. All of these message sites offer a comprehen sive online user transaction suite, which enables customers to conduct a full range of brokerage activities. An interruption in the messaging backbone can have serious ramifications for our clients or firms. As a result, service level agree ments (SLAs) between Thomson Reuters 50 and its clients or firms usually carry critical and short time requirements. Some SLAs also carry regulatory ramifications if the data in the message is inaccurate, incorrect, or not received on time. Keeping that informa tion flowing requires a coordinated effort among several groups within the Thomson Reuters BETA Systems organization, a part of the larger enterprise that serves the financial markets division. Some customers receive or process a million messages a day. Thomson Reuters BETA Systems follows an application lifecycle approach to ensure that applications are designed to meet custom ers’ needs and operate as planned, that any problems are resolved quickly, and that any lessons learned are communicated back to the innovation team for consideration in the next application rollout. Team members listen to the customers, determine how to deploy each customer’s application on the Thomson Reuters infrastructure environ ment, and then support the application. The backbone of this successful messaging environment entails three key areas: innova tion, creativity, and support. Innovative Solution Designs to Meet Business Requirements Each client or firm has different require ments and message volumes, so ever y messaging infrastructure that we design and create is unique. Most customers need a close-to-real-time messaging solution. Some customers receive or process a million messages a day; for others, it’s a few hun dred thousand. This is the innovation piece: The application development team translates each custo mer’s requirements into a unique solution design. The team determines the best way to move data from one application to ano ther, which often is determined by where the applications reside. If third parties are involved, then the application devel opers identify how the third parties are sending the information and where the information needs to go. The application development team decides which inter faces, middleware, and communication tools it will use to deliver information and solutions to customers. Creative Architecting for Success Next, the architectural team and the admin istrators of the middleware messaging team step in. This group is responsible for the middleware of the customer solutions; spe cifically, this group sets up the queues and channels. Middleware is the key to moving the information and transactions around different systems, which include UNIX, Windows, and z/OS. Middleware messaging occurs across both mainframe and distrib uted platforms that generate or process messages. Most of the messages, which primarily come from the mainframe, require a specific message format translation or a specific filtration. Creativity comes into the picture as this architec ture team ensures that client solutions are designed to be the most efficient method for each client or firm. The team answers questions such as: What type of messages will the messages be? What are the requirements for delivery? Who is it going to? Does Thomson Reuters have a preexisting communication to that client or firm? Is this a new client or firm? What should be set up for the monitoring? Is this something that has a time-frame requirement? Does it have low-latency messaging requirements? Or is it just messaging where everything is OK as long as the customer receives it within that specific day? For example, if the customer wants a specific message that delivers cash information, the middleware messaging team defines the steps necessary to complete that transac tion. The team identifies the message, the source of the message, the volume of the messages, the SLAs, and anything else needed to set up the proper support environ ment to align IT with the business needs of the client. Middleware is the key to moving the information and transactions around different systems. T his group ma na ges t he day-to - day operations through the solution rollout when business par tners or end-user clients are using the applications. The mid d lewa re a rc hite c t s a n d a d mini strators also form the second-line support. When a situation is not automated and cannot resolve itself, or when a problem is not a quick fix and an MQ administrator must troubleshoot, then this group is pulled back into the loop. 51 Close to Real-Time Delivery of Millions of Messages Effective Support to Maintain Availability and Performance of Customer Solutions support through a client service center and a technical operations department. After the rollout of customer solutions, the support team ensures that those applications remain up and running. Because every cus tomer solution is unique, our systems management must be flexible to accommo date each and every customer. Depending on a customer’s requirements and the SLAs, the middleware messaging department will implement different alerts and reporting. If a message needs to be somewhere within milliseconds, then the system alerts must notify everybody if a message has been sit ting in the queue longer than this SLA. Monitoring and Automation The monitoring system tells firstlevel support exactly what to do with each alert. We also make sure that the people respond ing to issues have the information they need when they need it and in the way that they need it — whether it’s on a page, a cell phone message, or a dashboard. We provide 24x7 We automate support as much as possible. We use software to monitor both the specific environment in regard to the messages and the environment that the messages are com ing from. That monitoring system sends alerts about anything that might create a problem with the sending and the receipt of the message to customers. The monitoring system tells first-level suppor t personnel exactly what to do with each alert. In some instances, the monitoring and alerting process is auto mated to indicate the steps required to resolve the issue. In other cases, the alert might indicate that the problem involves a specific customer and instruct first-level support personnel to forward the alert to the technical operations department if the problem requires specific expertise. The aler ts are automated according to the specific case-by-case scenario. 5 benefits of a Comprehensive Middleware Management Solution »» Proactive alerts of potential problem areas »» Faster problem resolution »» Single interface for monitoring and managing all platforms »» On-demand reporting for trending and analysis »» Increased ability to meet SLAs 52 Reporting We generate reports on the volume of mes sages per customer and use that data for billing. The organization also uses the analy sis to determine the time frames in which the messages went through. Because we sup port clients and firms in the financial services industry, targeting specific time frames is critical. Most important is the time between the opening and closing of the market at the New York Stock Exchange. When analyzing message delivery, the organization looks for a specific time frame on Mondays through Fridays, whenever the market is open. The monitoring and reporting capabilities help IT find and fix problems faster — and meet performance objectives. All this monitoring and reporting not only provides the necessary information about utilization, trends, and analysis for the IT organization, but it also provides our custo mers with the service level that they expect. The monitoring and reporting capabilities help IT find and fix problems faster — and meet performance objectives. Five Key Features of an Effective Middleware Management Solution When evaluating a middleware management solution, here are some key features to look for: »» Enterprise-wide middleware coverage of all the platforms that you have in use, and a single interface, reducing the com plexity of the solution »» Proactive notification of hot spots with the ability to quickly detect conditions that are slowing down key applications »» Specific problem identification and location to pinpoint both the specific problem and where it resides »» Wide variety of dashboards with rolebased access to enable users to quickly and easily find and fix the specific problem, contributing to lower mean time to resolu tion (MTTR) »» On-demand reports for trending and problem resolution, with easy access to historical reports that can be used for planning and identifying trends in capacity and performance Aligning Middleware with the Business At Thomson Reuters BETA Systems, every messaging infrastructure that the organi zat ion designs and creates is unique. By unders tanding a client’s requests, by being innovative and creative, and by establishing systems management that enables flexi bility, the organization can develop and support a unique messaging infrastructure that enables critical decision making for each client or firm. ● About the Author Felix Garcia is team lead for WebSphere Middleware Messaging and Automation at Thomson Reuters BETA Systems. 53 Flexible, Yet Reliable: Meeting the Challenges of Global Business Flexible, Yet Reliable: Meeting the Challenges of Global Business The needs of global IT users are growing in complexity and diversity. To meet these requirements, use a hybrid solution consisting of distributed systems integrated with a centralized mainframe. Distributed systems offer flexibility in meeting local user requirements, while the mainframe offers superior reliability and the benefits of centralized data management. By Bruce Chen 54 55 Flexible, Yet Reliable: Meeting the Challenges of Global Business In I.T. and in the world in general, physical walls are being removed. People travel more, and national borders are less important than ever before. More access to IT systems occurs via the Internet. And the number of sophisticated IT users who employ a wide range of client devices — for both business and personal use — to access IT systems has proliferated. Challenges of a Global Business Environment The modern global IT environment presents some unique challenges. Previously, most IT systems were localized to a single time zone or even a single local economy. Systems needed to be available during business hours, and maintenance could be conducted during off-peak hours. Now, business and retail IT users access IT systems from mul tiple time zones. IT is more likely to need 24x7 hours of operation, which changes maintenance requirements. And back-end systems need to be more stable to minimize unplanned downtime during all hours. Also, a growing diversity of end-user client environments in a multinational region such as Asia-Pacific creates significant differences in cross-culture, cross-border, and crosstechnology user maturity. Some users access information systems from traditional desktop client applications or through main frame terminal emulation. Others conduct business from mobile laptops or via the Internet. The use of handheld device applica tions that provide retail and business enterprise functions has also significantly increased. 56 Balancing Flexibility with Stability So, what do you do? On one hand, the front end of computing systems needs to be more flexible. You may need to localize different client/server applications to meet the needs of different countries. Or you may need to use different client applications for different plat forms that range from desktop computers, to store-based kiosks, to handheld devices. A structure of multiple distributed systems connected to a centralized mainframe is becoming a very popular system design. On the other hand, it is more important than ever to have stable back-end systems to maintain application functionality, transac tion integrity, and service quality. Supporting a diverse and distributed client environment requires integrating different distributed servers with a common back-end architec ture. As a result, a structure of multiple distributed systems connected to a central ized mainframe is becoming a very popular system design. The Resurgence of the Modern Mainframe Not too long ago, it seemed inevitable that the mainframe would eventually be replaced by distributed servers. However, the use of the two architectures has reached a balance point. For systems that have a very large, single database instance, significant I/O bandwidth requirements, intensive batchprocessing jobs, and quality-of-ser vice requirements, some workloads are now moving back to a mainframe. Across Asia, as in most of the world, end users are sophisticated technology consum ers and require more options for local operation. Many applications are tailored to a location by using specific application or deci sion logic. To simplif y deployment and maintenance while allowing scalability, many applications are written in Java or use ser vice-oriented architecture (SOA) technology. Applications may use MQ messaging or other middleware technology to access centralized data and transaction processing on a mainframe. It is difficult to maintain transaction quality in a distributed environment where distributed servers may span multiple locations, where client applications might be written in differ ent languages, and where middleware and systems management software are from a variety of vendors. So it makes sense for transactions initiated from a distributed environment to be executed at a central mainframe location. Centrally managing data in a geographically distributed computing environment offers several key benefits. Benefits of Centralized Processing Centrally managing data in a geographically distributed computing environment offers several key benefits. Distributed systems are effective for localized data input and output, but not for heavy computing and I/O. By centrally processing transactional data in a mainframe environment, you can greatly simplify the collection and analysis of data. Aggregated Reporting The most effective way to access data initially generated from multiple sources is to con solidate the data at a single location. If the data is in a centralized relational database, then you can easily access and query the data from multiple distributed sources. If, however, the requested data is stored in distributed systems and multiple databases, then queries and access based on a key or index is impractical. Because all the transaction data and service records are stored centrally on a mainframe, the batch processing of bills is greatly simplified. As an example, some telecommunications operators consolidate services from multiple distributed ser vice providers, such as Internet, local telephone, and long-distance telephone services. Cell phone operators also increasingly offer point-of-sale services that must be reconciled in monthly phone bills. Each service provider may have differ ent front-end applications that monitor and summarize billing data. To process a customer’s bill, all billing in- formation is consolidated to a centralized mainframe. If that data were collected from multiple distributed locations to process a bill, then large amounts of data query and transport would affect the performan ce of each distributed system. However, bec ause all the transac tion data and service records are stored centrally on a mainframe, the batch processing of bills is greatly simplified. 57 Flexible, Yet Reliable: Meeting the Challenges of Global Business Localized Tools Centralized transaction processing is an effective way to support localized applica tions. In another example, a bank in Taiwan is expanding and opening branch offices in many countries across the Asia-Pacific region. Different countries have different securit y rules, dif ferent per formance requirements, and in some cases, different maintenance windows. Each localized, frontend system must be modified and adapted to the requirements of each country. Central data management also offers data mining capabilities that create significant cross-sell and up-sell opportunities. without a mainframe to centralize the data and transaction processing. The bank maintains a central mainframe with a single database schema and transac tion engine. Each country may have specific data mapping requirements, but the transac tion logic and processing rules are common. The central mainframe has 20 different database management system (DBMS) engines. However, the central mainframe strategy requires only two database admin istrators. If each country had a distributed database, then maintenance and mapping requirements would demand significantly more resources. Customer Data Mining A user might initiate a transaction in one country that has a unique distributed envi ronment, but the transaction is processed centrally and then received in another country that has another variation of the distributed application. It wouldn’t be fea sib le to develop a network of distributed systems to support the transaction flow Central data management also offers data mining capabilities that create significant cross-sell and up-sell oppor tunities. Collecting and analyzing transaction history and behavioral data is problematic in a dis tributed systems environment. Data mining from a centralized database helps you target usage patterns that suggest opportunities for 5 Reasons to Centrally Process Transactions »» Better maintains transaction quality »» Simplifies the collection and analysis of data »» Enables aggregated reports from data generated at multiple locations »» Offers significant data mining capabilities »» Requires fewer IT staff to manage 58 offering products and services most likely to be of value to a particular user. Identifying associations between customers also helps facilitate referral business, which is a power ful way to drive new business opportunities. different languages in the different distributed systems — but also with the transaction identifiers and system names that may be in different languages. As a result, it isn’t always easy to integrate data from different systems. Hybrid Model Challenges When you build a centralized database, by design you will account for different coun tries, systems, names, and languages. If each locality defined what data should look like without having a centralized standard, then how would you manage the data? You can’t. It’s imperative that you have a centrally managed database so that you can manage the complexity. Although a hybrid mainframe and distributed architecture offers advantages over a single platform solution, the hybrid approach pres ents unique systems management challenges. With a centrally managed but distributed architecture, accessing and analyzing systems management data from multiple localized environments are issues when monitoring end-to-end transactions. Distributed systems may use monitoring and management tools from multiple vendors. Those tools may also have different native languages. As a result, consolidating and analyzing system messages can be prob lematic. In some cases, it is difficult to see different systems on the same screen. Scanning distributed systems can be a chal lenge if performance data is not in the same format or language. It’s imperative that you have a centrally managed database so that you can manage the complexity. For example, a transaction may start in Taiwan, go to a central mainframe, and then proceed to an office in China. The transaction and data may span multiple networks and servers that each may have different moni toring tools. The complexity is not only with business data — where the user’s name, address, and invoice number may be in Six Key Recommendations Six key recommendations can help IT organizations plan for global grow th using a hybrid mainframe and distributed system approach. 1. Use Distributed Servers When You Need to Localize Applications Distributed servers enable you to more easily customize and localize as needed to meet the computing requirements of the end user. Eventually, applications may lead back to a large database on a mainframe, but closer to the end user you need specific servers to meet their unique needs. In the Asia-Pacific region, for example, sepa rate localities and countries may have specific business needs and requirements that will differ from those of another locality. Those various needs and requirements are best served by a distributed environment, the server environment, because you can better customize. You need a combination of 59 Flexible, Yet Reliable: Meeting the Challenges of Global Business flexibility on the end-user level and security and performance standards at the back end, which involves the mainframe. 2. Leverage the Centralized Data on a Mainframe to Drive Incremental Revenue You might use distributed servers to address the immediate needs of localities. But having a centralized database on a mainframe — instead of separate databases in many locations — enables you to look at the overall transaction history across many regions and mine that data for revenue opportunities. Because of its scalability, performance, and ability to address security issues, the mainframe is still the best platform for centrally storing business-critical data. And having a centralized repository of cus tomer data makes it easy for you to drive increm ental revenue by using it to mine data and to identif y up-sell and crosssell opportunities. Because of its scalability, performance, and ability to address security issues, the mainframe is still the best platform for centrally storing business-critical data. is more stabilized. In contrast, country B is new to the finance world and offers plent y of new business oppor tunities. To maximize the profit in these two reg ions, the common strategy for country A is to increase the effort in taking care of existing customers and expend less effort on looking for new customers (upselling). However, the strategy should be the opposite for country B: Creating new opportunities and gaining market share should be higher priorities (cross-selling). Such business practices should be custo mized on the distributed servers, and the SOA should be designed according to business practices. 4. Ensure You Have the Ability to Manage Across Time Zones Generally, enterprises have more peak users during the hours when people tend to be awake. The Asia-Pacific region encompasses such a wide time-zone berth that you need full availability to provide users with access to their data. Invest in tools that allow your centralized database to maintain full per formance and 24x7 availability. 5. Measure the End-User Experience 3. Account for Regional Differences When you’re designing the centralized database, be sure to account for regional and language dif ferences in business data and IT systems. Not only that, the differe nce in business practices among regions should also be taken into consi deration. For example, let’s say one bank has busin ess in both country A and coun tr y B. In country A, the finance ind ustry has been developing for some time and 60 Even though much of your IT systems man agement is centralized, you need to have tools that allow you to understand and measure the end-user experience. For example, services in Japan and Australia might be running fine, but the end users in Korea might not be getting the business service that they require. When you run a data center and you have this type of hybrid environment, then you must manage your systems both centrally and locally. 6. Educate Others About the Value the Mainframe Provides As globalization increases and business needs become more diverse, it’s a good idea to remind the var ious busines s leaders of the strength of centralized processing. If your end users in Korea, for example, see that they’re getting the business service, they may not care about what’s happening in another country. But IT stakeholders and business leaders con tinually should be reminded of the benefits and the value that the mainframe — that is, centralized processing — provides while allowing them the flexibility to meet their localized needs. Invest in tools that allow your centralized database to maintain full performance and 24x7 availability. Don’t assume that your end users under stand the benefits of centralized processing, because at the end of the day, their needs are very localized. They need to understand why the mainframe is valuable and why it’s the best means to run their business, even though they have localized concerns. Hybrid Architecture Supports Global Growth As businesses go global and end users make more demands, using a mainframe as the back-end data and transaction server with localized distributed servers offers a far bet ter solution than either architecture alone. The connection to end users is becoming more sophisticated and encompassing a wider variety of client environments, making it more important than ever to have a main frame anchor for critical business services. The mainframe is not going away anytime soon. With a hybrid environment, the main frame can anchor a global growth strategy for the foreseeable future and beyond. ● About the Author Bruce Chen has been managing IT systems in the Asia-Pacific region for the past 25 years. He has more than 20 years of experience in data center management and has worked at a variety of different vendor sites throughout the years. Chen has helped to develop several successful enterprise applica tions that sold not only to domestic enterprises and government agencies, but also to international companies. He has an M.B.A. from National Taiwan University. 61 The Seven Habits of Highly Effective Mainframe Teams The Seven Habits of Highly Effective Mainframe Teams The mainframe has a bright future, but only if you continue to leverage your mainframe investment and explain how it maximizes the entire corporate IT investment. To accomplish this, practice the seven habits of highly effective mainframe teams — those teams with a track record of successfully leveraging and promoting their mainframes. By Bill Moran 62 T he mainframe is a vibrant processing platform; it has a bright future within the enterprise. Mainframe IT staff know that the mainframe offers superior perfor mance, availability, and security in processing critical computing workloads — both batch and online. They have watched the mainframe become ever more efficient and flexible, with the ongoing evolution of its architecture and the availability of Linux on the mainframe. Today’s mainframe is well equipped to oper ate as a highly cost-efficient server in the data center and to run advanced applications. Proactive collaboration with distributed systems teams can help expose a new generation of IT technicians to the mainframe and its modern capabilities. Unfortunately, to others in IT — and to senior management — the mainframe may be a mystery. Worse, it may be perceived as an expensive and outdated piece of hardware cranking away in the depths of the data center. Therefore, IT organizations still face a few challenges in getting the most from their mainframe investments. For example, they must be able to overcome organizational silos and increase recruitment. And they must be able to demonstrate that the main frame actually helps maximize the entire corporate IT investment. Meeting the Challenge Mainframe teams who have met these challen ges make use of seven critically important habits. Read on to learn about these habits of highly effective mainframe teams — along with suggestions for senior IT management — that can ensure a productive future for the mainframe in your organization. 1. Collaborate with Your Counterparts Your mainframe, UNIX, and Windows teams should constantly share information about the way things need to be done. This proactive collaboration (as opposed to the reactive collaboration in a war room) increases the overall productivity of IT and lowers costs. However, collaboration rarely happens out side of war rooms. Try to be proactive in your leadership: Cross-train and cross-manage your teams. Various advanced tools from vendors can help. For example, at one Ptak, Noel client, a new staffer was assigned to diagnose a mainframe problem using a GUI-based, cross-platform tool, and a main frame veteran was assigned a similar problem using a 3270-based traditional tool. The new staffer’s productivity nearly equaled the veteran’s. A large inhibitor to mainframe growth is its complexity, along with the fact that main frame talent is looking to retire. For the mainframe to continue, new staffers will need to enter the field, and it is important to recognize this reality. Proactive collaboration with distributed systems teams can help expose a new generation of IT technicians to the mainframe and its modern capabilities. Mainframe management tools are constantly improving, enabling non-mainframe IT staff to manage mainframe resources with little or no training. Mainframe tools are also evolving to appeal to a new generation of IT technicians. For example, event management 63 The Seven Habits of Highly Effective Mainframe Teams software can now send alerts to iPhone and iPad devices. how new mainframe applications could benefit the business. Although tools can help, staffers first must be willing and able to learn and use the tools. That’s where your leadership and inspiration come in, which leads to the next habit. 3. Actively Seek New Opportunities 2. Leverage People, Not Just Technology Remember that the knowledge in the minds of your mainframe experts is a valuable asset. They’ve probably been in the business for a while, and they really know how to run big, brawny applications 24x7x365. Also, they’ve been working with virtual systems for years. Virtual sprawl and virtual stall may appear to be new kinds of problems, but in fact, mainframe teams faced and solved these kinds of problems years ago. They have a lot to offer to distributed systems teams. Capture their knowledge and wisdom. Build it into processes, of course, but also use it to inspire and train newer hires. Everybody wins: The veterans receive recognition and satisfaction, and the newer hires quickly become valuable to the business and advance in their careers. Don’t hesitate to involve the mainframe experts in decision making. For example, most companies that are maintaining mainframe applications are unwilling to develop brand-new mainframe applications, usually because the mainframe experts have not been tapped for their expertise in Keep looking for places where the mainframe can efficiently solve business problems. For example, new Web applications that handle high transaction volumes are usually ideal. Here’s a great example of what can happen. The IT organization at a Ptak, Noel client reduced its expenses to levels of five years earlier, while increasing transaction rates several times. The company achieved this by migrating work to the mainframe. It had been migrating away from the mainframe, but in this case the company made an againstthe-grain decision. As a result, it achieved a dramatic improvement in its business. Keep looking for places where the mainframe can efficiently solve business problems. And many clients have consolidated multiple Linux systems on the mainframe and saved on software, air conditioning, power, and cooling. The mainframe can also save signifi cantly on staff costs. Roughly speaking, in the distributed systems environment, you need more staffers than would be required on a mainframe.1 This means the mainframe could, in many cases, provide the same — or better — service to the business units at a much lower cost. Take advantage of these more-favorable economics by identifying problematic 1 Of course, this depends on the mix of Linux, UNIX, and Windows in the distributed environment. You must make the comparison accurate by counting all support personnel. 64 applications with high service-level requirements and high operating costs. How can the mainframe help? Proactively identify opportunities where the mainframe could save money or improve business performance. In most organizations, there is a communi cations gap between the mainframe team and senior management. More so than the technology gap, this communications gap often makes it difficult for IT organizations to understand how the mainframe can solve “non-mainframe” problems and maximize the overall investment in IT. Take the initiative and bridge that gap by proactively identifying opportunities where the mainframe could save money or imp rove business perfor mance — then talk about those savings and improvements in business terms. 4. Think Outside the Mainframe While it was evolving from the sole processor to the multi-tiered environment, the main frame has changed in many different ways. The most conspicuous change is the user interface, from green screen to GUIs. It’s not just a cosmetic improvement; it’s also an important cultural improvement, because new staffers are no longer forced to learn an interface they may regard as obsolete. The mainframe, which started as a batchoriented system, has been extended to transaction processing and the Web — and it still does an outstanding job of batch processing. This dual capability enables the mainframe to play a flexible role in today’s hybrid data center. Architectural advancements such as zLinux are very popular, and the new zEnterprise System shows great promise. It has the potential to provide unified management of much of the data center. Even the development environment has evolved. You can now buy an attractively priced emulator that allows you to do zSeries development and initial testing on a PC. For many IT organizations, this is a major productivity boost. Use all this knowledge to benefit the whole business. Push yourself and your vendors for new ideas that stretch the mainframe’s utility. 5. Show the True Cost and Value Determine and demonstrate the true, accu rate cost and value of the mainframe, and make an apples-to-apples comparison with distributed systems. Remember the comm unications gap between the main frame team and management. The main frame team should take the initiative and present its numbers to all stakeholders. Start by reminding senior managers that they have already made the acquisition investment in the mainframe in many areas: the hardware, the staff, and the software. Point out that internal expense management systems may still have the mainframe paying the full freight for the data center, including floor space, electricity, heating, ventilation, cooling, and air conditioning. Try to develop an accurate picture of your IT expenses. 65 The Seven Habits of Highly Effective Mainframe Teams In some companies, even if senior manage ment agrees that the mainframe is cheaper, they still prefer distributed systems because it’s so easy to assign costs. Major vendors can help answer this objection; they have software that helps to assign mainframe costs easily. In addition, calculate the benefits. Don’t be timid; be aggressive. Remember the 80/20 rule: The mainframe is 20 percent of the over all expense but does 80 percent of the work. Don’t forget continuity and disaster recovery. If a mainframe is properly maintained, you won’t even have unscheduled downtimes. The cost of outages on your distributed systems is variable by business type, by applications, and so on; it’s difficult to deter mine the cost precisely. However, the cost tends to be a significant number over time because, when the distributed system with the database goes down, it may take many hours to get it back up, to rebuild the data base, and to correct it. 5 beneFits Mainframes are able to operate at a 90 per cent utilization rate, whereas distributed systems may operate at 20 or 30 percent. In other words, the mainframe provides a strategic advantage: a far greater ability to meet service level agreements (SLAs). 6. Talk Business, Not Technology Most IT leadership is concerned with meeting SLAs and controlling costs and may view mainframes as old-fashioned and expensive. When mainframe folks use “mainframespeak,” they confirm the belief that mainframe people are living in the past. The key for the mainframe team is to always talk in terms of business requirements. Start the conversation by observing that the main frame can help the business maximize its investment. Find out what your business counterparts need. How can you serve their business units in the best possible way? With its unrivaled availability, the mainframe can help deliver the level of service defined in SLAs. of Practicing the Seven Habits of Highly Effective Mainframe Teams »» Raise staff productivity »» Control costs »» Improve continuity »» Facilitate quicker disaster recovery »» Enable business to enter new markets 66 For example, is there a database that is un available and holding back some key busi ness objective? The mainframe could almost certainly help there. Here’s an examp le: A federal agency was using zLinux and wanted to serve other agencies. It talked about the Linux services it could provide and the guaranteed SLAs it was willing to sign. It didn’t mention the mainframe at all. This approach was quite successful. The lesson: Lead with the ser vice and it s benefits, not with the mainframe or its capacity. Don’t even mention platform; users may never know or care that they’re running on a mainframe. Lead with the service and its benefits, not with the mainframe or its capacity. 7. Be a Reasonable Champion Be careful not to overdo your advocacy. By all means, the mainframe is a very effective platform, so take credit for the mainframe where credit is due. But don’t risk your cred ibility by claiming that the mainframe is the ideal platform for everything. It’s not. It’s OK to acknowledge that for certain applica tions, UNIX, Windows, or Linux stand-alone systems might be the best solution. In the Web-based environment, the main frame plays, or can play, a significant role. It offers outstanding availability, the best of any general-purpose system you can get today. For some businesses, the mainframe is the place to put the front end of the Web, but in many businesses a distributed system might be better. Utopia will have arrived when the mainframe is so smoothly integrated into corporate IT that people will forget they’re using it. They’ll just know they’re getting great service. Maximize Your Investment Ptak, Noel clients who practice these seven habits of highly effective mainframe teams gain substantial business benefits. Your organi zation can, too. You can raise staff productivity, control costs, improve continuity and disaster recovery, and enable business units to enter new markets. Your mainframe will help max imize your entire corporate IT investment, and you will have the numbers to prove it. ● About the Author As research director at Ptak, Noel & Associates, Bill Moran leverages his more than 30 years of experience in IT analysis, technical management, and technical education to help clients elevate their solution sales discussions out of the technical weeds. He also organizes technical updates for clients on such issues as new hardware, new software, and implications of competitive actions. Moran has also held research positions and senior management positions at Ideas International, D.H. Brown Associates, and Coopers & Lybrand. Previously, he held management and technical management positions at IBM and was responsible for the company’s worldwide, large-systems support for the IBM World Trade Corporation and benchmark testing for IBM’s mainframe systems. 67 Ten Tips for Mainframe Success Ten Tips for Mainframe Success IT organizations can reduce and optimize mainframe costs by taking advantage of specialty processors, selecting the right vendors, and continually tuning performance. Follow these ten tips to make the most of your mainframe investment. By Jonathan Adams 68 69 Ten Tips for Mainframe Success W hat’s top of mind for IT leaders today? In response to BMC Softw are’s 2010 Annual Main frame Survey, 65 percent of the respondents indicated that reducing IT costs was a top priority. And this is no surprise; not only was this the most important priority indicated in the survey, but it is what we hear continually in visit s to customer sites and from exec utives who are mem bers of the BMC Mainframe Advisor y Council. IT organiz at ions from around the world are saying, “Help us reduce costs to our business and make the most of our IT investment.” A comprehensive workload automation solution can help you effectively redistribute workloads without impacting the business services that they support. What can you do right now to reduce costs and get the best value from your existing investment in the mainframe? Focusing on cost optimization is a good s t ar t . C os t optimization is all a b out maxim izing the value of your investm ent in the IT infrastructure. But first, you need to know which factors are driving most of your costs. And in almost every case, cost is associated with capacity or MIPS utilization. You can reduce and optimize mainframe costs by concentrating on three main areas: Take advantage of specialty processors, select the right vendors to work with, and continually tune performance. 70 Tips for Maximizing Your Mainframe Investment Here are ten tips to effectively achieve cost optimization for the mainframe within the three areas mentioned previously: 1. Use zIIPs and Specialty Processors to Reduce Costs One of the most exciting things that IBM has released recently is the assist processor, a zIIP or zAAP engine. These are considerably less expensive processors than generalpurpose engines and can run specific types of workloads so that you can offload those workloads from a general-purpose engine. In addition, with a general-purpose engine, after you purchase the hardware, you still need to pay software license charges. With a zIIP engine, there are no costs beyond the initial hardware. But even though the overall cost of a zIIP engine is lower, you still want to make the most of your investment. Look for vendors that offer products that have zIIP enablement. For example, tools should enable you to easily identify zIIP-eligible workloads so you know what workloads you could offload to a zIIP processor if you were to buy it. Make sure you understand the impact of zIIP and specialty processors. 2. Identify Ways to Lower Your Four-Hour Rolling Average of MIPS Utilization Some outsourcing vendors will charge you based on your four-hour rolling average of MIPS utilization. This is another area where you need to understand the load on your machine and tie it back to the contract with the vendor. How can you reduce that four-hour rolling average? Moving specific workloads to specialty engines can definitely help. You can also redistribute the timing of your workloads to move some of them out of peak processing periods. A comprehensive workload automation solution can help you ef fectively redistribute workloads without impacting the business services that they support. 3. Don’t Save Money at the Expense of Service Levels Look for software solutions that can model what workloads will look like on a specialty engine. It’s not enough to just move them from your general-purpose engines. Ask yourself: Do I still have the desired response time? Do I still have access to the direct access storage device (DASD)? Is there anything that might make using the assist processor not so beneficial? Be sure to identify the workloads and model them so you know the end result before you start. You don’t want to save a few dollars off the bottom line only to lower your service levels to the extent that you cannot meet your business requirements. Well-written code can accomplish the same amount of work with less MIPS utilization and less resource utilization overall. 4. Realize That Efficient Code Is King Look for solutions providers that have effi cient code that can perform the operation or workload effectively in the least amount of time, not only from an elapsed time per spective, but also from a CPU perspective. Well-written code may run two or three times as efficiently as poorly written code. And well-written code can accomplish the same amount of work with less MIPS utiliza tion and less resource utilization overall. A good solutions provider will benchmark its own products, release over release, and constantly strive for improvement. If anything changes in the environment, such as a new operating system or a new version of a sub system, the solutions provider should always be looking for ways to improve the efficiency, effectiveness, and speed of its products. All of this translates to cost savings. 5. Look for Vendors That Can Offload Some of Their Own Work to Specialty Processors An effective vendor tool will identify and model what you can offload to a specialty processor, and actually help you by putting some of its own processing onto the zIIP engine. That’s another way a good vendor can pass off savings to a customer. For example, if one piece of code is more efficient than another and runs with less of a footprint on the general-purpose engine, then go fur ther and move a portion of that footprint to the less-expensive engine, freeing up even more general-purpose MIPS. 6. Evaluate Whether Outsourcing Is Right for Your Business If you are looking at outsourcing, be sure that you can do so without losing a competitive advantage. When resources are under your own roof and your internal staff is looking after them, you have a certain degree of control. But you also take on the pain and the burden of managing everything, so it’s unders tandable why you might consider outsourcing. However, be sure outsourcing is right for your business. 71 Ten Tips for Mainframe Success Some companies have determined that IT is such an integral part of their business that they need to have IT in-sourced and wholly owned to compete effectively. For example, at a large U.S. bank, the CEO decided the company could not differentiate itself from its competitors if it had a commoditized IT environment. He believed that banking relies so heavily on IT that his bank needed to own that function to be differentiated. In other words, they had to do IT better than the competition: The banking products and business applications they put on the market had to be better and hit the market faster than everyone else’s. So, he chose to go the in-sourcing route. 7. If You Decide to Outsource, Carefully Conduct Due Diligence on the Outsourcing Vendor If you determine that outsourcing is right for your business, make sure you implement the right levers and controls to maintain the service level agreements you have with your customers in terms of responsiveness, speed, and transaction times and volume. In addition, look for confirmation that the vendor is acting in your best interest and is striving to optimize the resources used. If you determine that outsourcing is right for your business, make sure you implement the right levers and controls to maintain the service level agreements you have with your customers. For example, if there is a problem with re sponse times, be sure the vendor is not adding huge amounts of capacity to fix the problem and then charging you more than you should be paying for capacity — especially when the vendor could use the hardware more efficiently. You can address this by having good controls in the contract, with a periodic rationalization to net out how much capacity you used versus what is in the contract, efficiency studies, and throughput studies from time to time. Everyone understands that the outsourcer needs to make a profit to stay in business, but a good outsourcer will focus on running as efficiently as possible so that it can offer a fair price and yet still make a good margin. 5 points to Consider When Evaluating Mainframe Value »» Remember that the lowest bid does not automatically mean the best value. »» Recognize that price is not an indicator of value. »» Analyze efficiency, reliability, and performance. »» Consider security and risk. »» Calculate the total cost of ownership (TCO). 72 And be sure you’re not stuck with a contract that doesn’t allow your business to respond to changing needs three years from now. Price is not an indicator of value, and the initial price tag shouldn’t be the only consideration. 8. Understand That ISVs Provide Good Value for Your Investment In the past, the perception of the high cost of independent software vendor (ISV) software has been cited as an inhibitor to mainframe growth, preventing IT organizations from putting more projects and applications on the mainframe. Today, savvy IT organizations are realizing that ISV software can help reduce total costs by lowering MIPS footprints through the exploitation of more efficient code and utilization of lower-cost processors (zIIPs). In other words, you can save money over the long haul with more efficient sof tware that accomplishes the same amount of work using a smaller footprint of hardware, DASD, resources, and attention from the IT staff. So look at all the vendors, line them up, and benchmark them to see who’s more efficient. The 2010 mainframe sur vey shows that companies are now saying that ISV software is no longer the big expense. Even if you are outsourcing IT services, you can retain the right to select and even contract for products directly with ISVs. In this way, you can better establish and maintain the most efficient mix of sof t w are products to meet your busi ness goals, such as availability, speed, and ef ficienc y. 9. Evaluate Long-Term Value: Lowest Bidder Does Not Always Equal Best Value Price is not an indicator of value, and the initial price tag shouldn’t be the only consideration. You’ll also want to consider long-t erm efficiency, reliability, and per formance; flexible terms and conditions; pricing options; and the two- or three-year total cost of ownership. Look for a vendor that can provide metrics and tie software to its actual utilization. By looking at the bigger picture, you can evaluate which vendor delivers the best value over the life of the contract. Mainframe cost optimization is the overall package; it’s not just about buying inexpensive software and hardware. The bottom line is long-term val ue. Buying the cheapest products may end up introducing more risk to your business. Finally, look to trusted vendors that have been in the market for a long time, who will be there in a time of crisis. You want a vendor who will stand behind you and will also be there to help get you back on a solid footing if the need arises. By looking at the bigger picture, you can evaluate which vendor delivers the best value over the life of the contract. 10. Look for Tools That Help with Dynamic SQL Tuning and Performance Select tools that can help reduce the applica tion footprint or resource consumption of DB2-related applications. Tune your in-house applications and make sure that they’re running properly. Also ensure that the per formance of your overall machine is optimal. 73 Ten Tips for Mainframe Success Closing Thoughts In all of this, remember that IT exists to serve the business. Make sure that you’re getting the most value for your IT investment, and don’t lose sight of the fact that you still must be competitive at the end of the day. If the business need requires you to be quicker to market and more efficient than your competi tors, you may choose to own rather than outsource your IT environment. But whether you in-source versus outsource, you need to take ownership of your costs. So take advan tage of specialty processors, carefully evaluate vendors and outsourcers, and con tinually tune performance, and you’ll be well on your way to getting the most out of your mainframe investment. ● About the Author Jonathan Adams is the vice president of Data & Performance Management of Mainframe Service Management at BMC Software. In this role, Adams leads the development, support, and quality assurance for the DB2, IMS, MainView, and Middleware Management products within BMC Software’s Mainframe Service Management Business Unit. Adams has been at BMC since 1995. He has worked on mainframe optimization and networking products, helped launch BMC’s enterprise-wide MQSeries mission in 1997, and has worked in the MainView and MSM organizations in various management positions since 2000. Prior to joining BMC Software, Adams worked in IT at BellSouth Services where he held a variety of analyst and systems programmer positions in technical support and data center operations. Adams received his bachelor’s degree from the University of Alabama and a Master of Business Administration from the University of Alabama at Birmingham. 74 A Compilation of Articles by Industry Experts VIEWPOINT Focus on: Mainframe Cost Optimization Focus on: Mainframe Cost Optimization The Mainframe Strikes Back: 5 Reasons to Stay with the Mainframe This publication was created by BMC Software. Business Runs on IT. IT Runs on BMC Software. The Keys to Effective Integration *199997* VOLUME 10 Business thrives when IT runs smarter, faster and stronger. That’s why the most demanding IT organizations in the world rely on BMC Software across distributed, mainframe, virtual and cloud environments. Recognized as the leader in Business Service Management, BMC offers a comprehensive approach and unified platform that helps IT organizations cut cost, reduce risk and drive business profit. For the four fiscal quarters ended December 31, 2010, BMC revenue was approximately $2 billion. Visit www.bmc.com for more information. The Economics of Computing 4 Guidelines for Maximizing the Value of the Mainframe Published by BMC Software