The Dinosaur Myth

Recent changes in the market have made the mainframe more attractive than ever, compelling one analyst firm to publish a new version of a past report.

Looking for something hefty to smack down on the boss's desk the next time he or she starts talking about moving more of the datacenter to open systems or even—gulp—Windows? Then check out the new numbers in a recently re-issued report from Xephon, a British analyst firm that covers the large information systems market.

Xephon has been researching the cost of ownership of large systems and software since the mid-1980s. Its 25-page report, titled "The Dinosaur Myth—Why the Mainframe is the Cheapest Solution for Most Organizations," was first published in 1992, when the idea of downsizing was as hot—ironically—as server consolidation is today.

Back then, moving from big, expensive centralized systems to smaller, cheaper, distributed platforms seemed logical and inevitable. Pundit Stewart Alsop's well-known snipe that "the last mainframe would be turned off in 1996" can now be matched by paraphrasing Mark Twain—that reports of the mainframe's death have been greatly exaggerated.

Today, most companies agree that mainframes aren't going anywhere any time soon. In fact, as Steve Swoyer discusses in this month's cover story ("The Evolution of High-End Servers,"), the mainframe is very much alive and well, thank you. Among other things, IBM's clever moves with its smaller "entry-level" zSeries mainframe, lower pricing, and its embrace of Linux are keeping the biggest systems viable and attractive. Sun, HP and Unisys, in particular, are chasing the high-end market with their own entries, either open systems or—in Unisys' case—as a partnership with Microsoft.

According to Xephon, recent changes in the market have made the mainframe more attractive than ever and compelled them to publish a new version of their report.

One interesting point the report makes early on is that cost of MIPS may not be the best way to compare systems—although it's commonly used in the industry. Instead, since much of today's computing is data-intensive rather than processor-intensive, a better measurement would be how a computer handles data—something mainframes have historically been very good at.

The report also contains lots of great quotes that you can toss out at your next purchasing committee meeting or in your next report to management on datacenter costs. Some samples: "Mainframes offer the most cost-effective computing facilities for all but the smallest organizations" and "Mainframes are the undisputed masters of batch processing." Here's one you probably won't be hearing from Microsoft any time soon: "A fair comparison [of computing costs] should take account of time wasted because a system is slower to respond."

You can read "The Dinosaur Myth" free online by going to, or purchase and download a reasonably priced PDF copy for the boss's desk, complete with a clearly irritated Tyrannosaurus Rex on the cover.

What hardware and software will your datacenter be running in five years? Write me at

About the Author

Linda Briggs is the founding editor of MCP Magazine and the former senior editorial director of 101communications. In between world travels, she's a freelance technology writer based in San Diego, Calif.