By the time you read this, the late 2009 version of the annual IBM Systems Technology Group (STG) analyst briefing will be a distant memory. Much of the messaging at this briefing was similar to years past. Cost containment? Been there. Cloud computing? Not too far from the hype about Software as a Service (SaaS) from five to 10 years ago. But there was one new theme that dominated the discussion: mainframe strategies.
Now, to be fair, IBM strived as always to give each platform its due. System x and System p received due attention as important ways for PC and Linux users, respectively, to scale cost-effectively and consolidate with energy efficiency. But when you probe beneath the surface of the overall IBM message, certain marketing themes repeatedly surfaced.
The importance of having as many virtual machines as possible on the same node? A mainframe TCO message stated over the past three years. The idea of “hybrid” computing, in which each platform specializes in a particular type of computing, and small servers are lashed to large ones, with workload flowing between them for load balancing? That goes back to the revised mainframe idea of the “hub,” introduced two years ago. Energy savings via physical consolidation? Mainframe TCO studies? Those ideas are from two to three years ago. Differentiation from other hardware vendors by scale-up performance, robustness, and security? That mainframe message has been around for a long time.
The key to understanding IBM’s renewed emphasis on the mainframe is competition with HP. So far, Oracle’s acquisition of Sun has been viewed as steadily decreasing the competitive hardware threat from Sun. While Microsoft has placed some emphasis on moving workloads off the mainframe and onto Windows, there are no clear signs that net outflow from the mainframe has increased. Although user interest in migration from z/OS has recently spiked upward, IBM noted that new orders of Linux-only mainframes had begun to make a nice little market.
If, in effect, the hardware competition has become a two-horse race, then IBM knows there’s one place HP won’t go in the near future: scale-up. Too much emphasis on scale-up will undercut HP’s strengths in scale-out Wintel and Lintel; nor are VMWare instances on a single Lintel box likely to scale to the hundreds in the near future. If HP should go there, Fear, Uncertainty, and Doubt (FUD) about HP’s ability to deliver mainframe-like robustness and security along with the scale-up would continue to differentiate HP and IBM for years to come.
Does this mean IBM is ceding scale-out to HP? Not at all; but if by some unhappy chance HP wins the marketing race there, IBM is placing a smart bet that scale-up will increase in attractiveness compared to scale-out. There’s no doubt that scale-out is relatively energy inefficient, and there’s a strong case that administrative costs for scale-out are higher, what with network administration and the like. Mainframe-type scale-up will be vital in the effort to cut computing energy expenditures because scale-out by adding boxes or invoking existing remote boxes is effectively “flow of energy to the least regulated source,” and is therefore guaranteed to have minimal impact.
The one area where IBM falls down in this hybrid-platforms-and-mainframe-differentiation strategy is the ability of Windows workloads to flow to the mainframe. As previously noted, third-party vendors are beginning to tackle the problem. There was no hint at STG that IBM yet recognized the scope of the Windows problem, nor the opportunities to be had by solving it. However, clearly, vendors such as CA are pressing ahead with mainframe administration tools—Windows-like user interfaces for replacing the retiring Job Control Language (JCL) experts—and IBM would dearly like to put a final “done” stamp on a cross-platform, cross-operating-system global set of administrative, development, and middleware software for x, z, and p.
My net-net for mainframe users: IBM is looking for you to help drive major enterprises toward scale-up. Pressure IBM about Windows; look at IBM internal clouds as a place to experiment with cross-platform workload shifting that involves mainframes; and encourage the business to focus more on going green in IT. You don’t need to be a mainframe advocate. If IBM and third-party software suppliers succeed, growth for every enterprise strategy will mean growth of mainframe MIPS. This year, if the creek don’t rise, the mainframe will be at the end of every strategic path, just as it was at IBM’s STG event.