Scalability: The Most Important Metric

By Charles Pelton

ou can buy all the IT infrastructure you want, but if the systems and processes that infrastructure supports can't contract and expand, then your company can't respond to market conditions.

A merger in the offing? Your organization must grow and adapt. A popular new product line? The hardware and software better be able to grow with that market. Look at the big picture. With five years of unabated growth in the U.S., a 15-year bull stock market, and a healthy pickup in the formerly sluggish economies in much of the rest of the developed world, the operative IT function better be scalability. After all, this expansion can't last forever.

Indeed, scalability has long been the advertising domain of large and m idrange system vendors. Even today, if you buy a smaller AS/400, you can be assured that it'll be (fairly) painless to swap it for a more powerful machine. Ditto when it comes to many Unix machines -- and of course MVS or S/390 architectures. If a portion of the premise of IT "upgrading" is embracing new technologies, surely the other reason always rests on the need to enhance, grow, and multiply business opportunities.

What about newer technologies? Why is there still so much resistance among IT executives to embracing Microsoft Windows NT? You always hear the objection about the lack of maturity in the operating environment. But at the end of the day, it always seems to come down to scalability.

At an InformationWeek -sponsored roundtable in New York last month a group of enterprise server and mainframe vendors gathered with a half dozen CIOs and a like number of IW editors to discuss the future of the mainframe. The discussion was seeded by a study of 250 read ers which asked, among other things, which was their operating environment of choice for enterprise, mission-critical systems. Some 63% said MVS, 16% said Unix, and only 6.4% said Windows NT. When asked to project ahead two years, the numbers shifted slightly in NT's favor -- 57% for MVS, 20% for Unix, and a doubling of the NT numbers to 12.8%.

What's holding back universal acceptance of NT? Or, better yet, is there anything more scalable than MVS? When asked if MVS is sufficiently scalable, 63% said yes. For NT, the number plummeted to 30%. But there's a twist. When asked to project out two years, nearly 54% said that NT will be sufficiently scalable.

NT -- or any operating environment for that matter -- will be embraced when it's viewed as a vehicle to succeed, not an impediment to growth. And success in the IT world may mean the ability to scale up to embrace new opportunities or even downsize to change market focus.

All this leads me to consider two more IT phenomena: the Internet and the mi croprocessor. One has enjoyed relatively recent acceptance and popularity; the other has been viewed, essentially, as an ever-more-powerful IT utility.

Is there anything more scalable than the Internet? Futurist George Gilder talks about the coming apocalypse of infrastructure and bandwidth. But no one is acting like there's any constraint on networking resources. The fallacy of the commons aside (and it will catch up with us all!), IT organizations are behaving as if the Internet offers unlimited communications and commerce opportunities. No wonder so many companies are flocking to Java, intranets, and browsers.

But if Microsoft, and by extension many other companies, push and embrace NT as a corporate standard, is the standard Intel microprocessor up to the task? During an InformationWeek visit in early August to Intel facilities in and around Portland, John Miner, VP and General Manager of Intel's Enterprise Server Group, hinted at an evolving enterprise orientation in the way his company develops microprocessors and related products. It's almost as if Intel is turning its own world upside down, looking at corporate needs first, then considering what powers faster and bigger PCs.

Says Miner, "Scalability in the future will be defined as just adding another server." And he adds that he'd "never like to be gated by an OS." Intel has ongoing initiatives and research partnerships anywhere there is an impediment to scalability -- the I/O, transaction processing technology, software, and clustering.

Admittedly, there are limits to growth on the PC platform. That's where NT's future scalability comes in. And that's why clustering -- really a scalability strategy -- is so important to the future of so many companies.

It has to be important. No company can afford to stand still. Few remain in protected industries. Business growth and IT scalability are inexorably entwined. A company with systems that can't scale will be forced to stop, evaluate its environment, and regroup. A company with systems th at can scale will be armed for all eventualities.

AuthorITies Archive

Send Us Your Feedback

Rich Levin: Run Time Rich fills you in on all of the latest products, issues, and trends in application development.		Stuart J. Johnston: Red mond Watch As our eyes and ears in Redmond, Stuart gives his perspective on the latest events at Microsoft.		Charles Pelton: Eye On I.T. Charles explores IT management issues and strategies that business and technology managers face.

Top of the Page