A couple of weeks ago, HP released a survey conducted by PSB Research that collected input from 490 business managers, CIOs, and IT managers around the world about their IT strategies. Here’s one of the findings emphasized in its accompanying press release: “[F]or one out of five technology executives in the study, moving away from mainframes is more likely, with the main motivator globally being the high cost associated with mainframe maintenance and licensing.”
Interested as I am in the recommitment to mainframes that I’m finding in just about every company I visit, and with survey data fresh in mind from a late-2008 report compiled by a different research firm—a study that showed a significant increase in the workload from distributed systems that was being migrated onto big iron—I was flabbergasted by HP’s claim.
Overall, the HP-funded study wasn’t entirely off base. It seemed to capture the same sentiments I was hearing in my travels. Cost-containment was driving companies to consolidate servers and storage, to modernize applications, and to build additional automation to improve efficiency and reduce staff. The conclusion drawn about mainframes, however, seemed to defy my sense of things as well as the stated goals documented by the study.
With mainframe workloads increasing as companies consolidate applications and their workload onto this platform, the goals stated by HP interviewees looked as though they were being well-served—elegantly and efficiently. So, why did “one in five” seek to de-leverage their mainframe investment?
I dug deeper into the survey, and it turned out that, of the 490 interviewees, 136 respondents were front-office (business) folks, while the balance were technology executives. Moreover, 75 percent were “enterprise” shops that might well have mainframes today, while the balance were “mid-market” companies that might be x86 systems-centric. There were no explanations provided regarding what designations enterprise or mid-market meant.
Going further, 153 interviews were from U.S. companies, and 133 were from European and Middle Eastern firms (pretty broad swath). Asia Pacific accounted for 128 responses, and Latin America 76. The survey company said their data might be off by four points, plus they conceded that results may be further skewed by “language differences” among non-English speaking respondents.
Given the demographics of the research, I wondered if all the interviews were conducted with companies that currently use HP. More to the point, I wondered how the numbers broke out in the different geographies under study.
Interestingly, the overall numbers painted a very different story than suggested in the HP headline: 70 respondents said they were staying the course with mainframes and committing even more to the platform, while 21 said they were less inclined to stay with the platform; 10 respondents had no answer. Only 13 of 354 respondents said they were strongly inclined to move away from the mainframe. That’s something like one out of 27, not one out of five, which would require 70.8 out of 354 technical respondents (the subset of respondents quoted on the mainframe question) to say they were abandoning the mainframe. This wasn’t the case.
By geography, North American respondents numbered 102, of which 14 were strongly inclined to move away from mainframes. In Europe and the Middle East, only two of 101 respondents were so inclined. Asia Pacific (101 respondents) and Latin America (50) showed the greatest percentiles to do so with 20 opting out of mainframes in the Pacific region, and 18 preferring this approach in Latin America.
Overall, the sample size and demographics seem too unscientific to make the claims proffered by HP reasonable or meaningful. The troubling thing is, however, that the summary statements, which defy all mathematical reasoning, may well become headlines in prominent trade press journals, on slides presented at conferences and trade shows, and in analyst prognostications.
What’s worse, the claim that one in five are abandoning the mainframe may find its way into common parlance and even into business publications, where it can influence the keepers of the corporate purse. A lie told often enough, so the old saw goes, often obtains the status of “common wisdom.”
At the end of the day, the hard-fought arguments in favor of the mainframe—the ones we can empirically support with cost-savings, risk reduction, and productivity improvement data—are vulnerable. If you don’t believe me, ask the tape industry.