A major open systems server vendor recently released a survey showing 21% of technology executives reported they are considering moving off their mainframe due to high costs during tough economic times. Contributing to the high cost reportedly were high maintenance and software licensing fees in a mainframe environment.
The survey respondents felt that open systems servers would give them increased flexibility and save money, time, and space while still addressing the needs of their current IT environment. What data were these execs basing their responses on?
Mainframes aren’t cheap. IBM, however, even made headlines back when it introduced the System z9 business class machines and touted a new low price. Since then it introduced the beefier System z10,promoting its even better price/performance. The IBM announcement includes the usual hype, but mainframe buyers tell me that the essential points are valid.
Starting a bit above $100,000 today the System z business class machines certainly aren’t low cost by typical open systems server standards. Probably, when all is said and done, you won’t get even the smallest up and running for twice that.
OK, I’ll be the last guy to defend IBM’s mainframe pricing although the company has cut prices and played with packaging and licensing terms to effectively reduce software costs. Still, the 21% of executives who are considering ditching their mainframe either made the most simplistic cost analysis or failed to understand what they had in the mainframe or what they would be buying.
A simplistic TCO analysis looks at the acquisition costs of servers and software and, maybe, the initial setup required to get them running. Such an analysis, however, fails to take into consideration all the other elements of the IT environment necessary for achieving the expected service levels.
There will be workloads and situations where open systems servers indeed offer a better deal. But finding them requires a comprehensive TCO analysis that looks beyond the initial hardware/software acquisition costs to examine the cost of:
- Redundant everything for purposes of high availability (99.99% or 99.999%) to meet service level objectives
- Scalability for when economic growth resumes
- Interoperability with middleware, networks, and the upstream and downstream IT infrastructure
- Utilization (the ability to maximize the capacity purchased)
- Labor for ongoing administration and management
This last point, labor, is often overlooked but critical. Various published studies suggest that the cost of managing IT systems over time runs 3-8 times more than the acquisition cost.
Over the years I have done ROI and TCO analyses for numerous vendors. It is possible to skew the results by low-balling various assumptions, especially labor, and selectively omitting related infrastructure costs.
To get a true picture of the value of a mainframe compared to the equivalent processing resources of an open system server you need to perform a comprehensive TCO analysis. And even then, make sure to consider expected levels of service now and going forward and the likelihood the organization will grow in the future.
Given all that, I’m not convinced 21% of those execs surveyed actually will ditch their mainframes anytime soon. And some of those that do may regret it.