There’s no single definition for Total Cost of Ownership (TCO) since every enterprise defines it relative to its particular business model, often borrowing from several vendors. Of course, every vendor presents sites with TCO formulas that favor their solutions. But if you’re a mainframe site, or one of the 23 non-mainframe shops that opted for a mainframe solution over the past year, labor-saving virtualization capabilities, coupled with attractive pricing on specialty engines, will be a big part of the mainframe value proposition.
“There’s a common trend that we see across all enterprises, which is the movement toward virtualization and server consolidation,” says Jim Porell, IBM Distinguished Engineer and director of Business Development. “System z is very attractive in this scenario, because it allows businesses with hundreds of Windows and Linux computers to gain as much as a 10:1 reduction in units; so, for example, an enterprise can conceivably reduce 800 servers to 80 by virtualizing on System z. The savings can be spectacular, and it’s making a lot of businesses happy.”
However, this doesn’t mean that it’s easy.
A labor savings value often ignites politics between mainframe and open systems groups, especially in a time of economic downturn. System gains in capability and automation are great, but corresponding staff reductions for the work saved aren’t so great for those impacted.
When mainframe TCO is computed and IT begins to see the savings in virtualization that eliminates server sprawl, avoids idle time in distributed servers, delivers more systems management automation and reduces the manpower needed to support distributed servers, the mainframe value proposition begins to crystallize—as does the potential impact on open systems jobs since manual workload is reduced.
“You can go look at any data center and see five people per shift working on the mainframe and hundreds of more staff committed to running open, distributed systems,” says Susan Eustis, consultant and principal analyst with WinterGreen Research. “This manpower disparity occurs because the mainframe runs a shared workload with many automated processes. You might have 3,000 applications on a single System z as compared to 3,000 applications on 15,000 distributed servers. When you evaluate TCO, you want to look at labor costs, because labor is 70 percent of the cost of running IT. That’s the funny thing about TCO; many companies develop a TCO formula that totally leaves out labor.”
Working on the labor part of TCO is uncomfortable for many IT organizations because it involves potential culture clashes—and job loss—within the organization. Many sites have addressed the issue by transitioning open systems support personnel to the mainframe, where their backgrounds in Linux and UNIX are quickly adapted to the world of z/VM—and accepting the fact that some staff will pursue employment elsewhere.
“It’s a scary prospect at first,” acknowledges one Midwest CIO. “But by spending the time with people as they transition into new job functions, they eventually see the logic of it and they adapt. Most important, they don’t see a big difference in applying what they already know in the distributed world to Linux on System z.”
Eustis believes there’s a natural tendency for many IT managers to think that open, distributed servers offer better TCO than mainframes because of the advice IT often receives from outside consultants who favor data centers with open, distributed systems. Many of these consultants have limited mainframe experience, so they aren’t well-positioned to understand what the mainframe can offer in TCO. Others are caught up with the lower “in the door” costs of deploying commodity Wintel servers. The result can be server sprawl and challenges in managing the enterprise IT payload data, applications, security, utilities, networks, and servers.