A recent report from McKinsey, based on data from the Uptime Institute, shows just how significant the issue of data center efficiency is likely to be in the years ahead. McKinsey makes the mind-numbing forecast that, by 2020, the data centers of the world will be pumping more greenhouse gases into the environment than the airline industry. Of course, it’s difficult to predict how air traffic will change over the next 12 years or how travel demands will evolve. But it’s a pretty safe bet that the growth in online data and applications will continue unabated during the same period, and the decisions we make now about the best platform for this growth will have some very serious implications for the future.
The jury is still out about whether the System z architecture is the best way to improve data center efficiencies and reduce the environmental impact of IT. Certainly IBM sees a big opportunity to sell the environmental benefits of consolidation and virtualization in this space, and many of the recent z10 announcements have a distinctly green focus.
Through Know-How, mainframers can get all the latest in news in the world of the mainframe. Know-How is completely free, and you’ll get access to content that includes thought-leading articles, whitepapers, webcasts, seminars, and information that will help you get the most out of the mainframe equipment and software you already own.
Ironically, one of the ways distributed systems, based on commodity processors, have been able to compete with large systems technology over the years has been by throwing low-cost hardware at any performance problem. When hardware was cheap and power was barely an issue, raw capacity increases were an ideal way to address the problem of performance management without piling on the costs. Now that power consumption and heat generation are troubling data center managers, the length and breadth of the land, the idea of centralized mainframe processing has become attractive once again.
But more than mainframe technology, mainframe-related skills, such as capacity planning, which are barely understood outside the mainframe world, are now becoming essential for companies that want to avoid the environmental and financial cost of the uncontrolled sprawl of server farms. The future of the data center might come down to reapplying large systems disciplines to other platforms, at a time when those skills are rapidly disappearing!
The cost-effectiveness of IBM’s new z10 mainframe is standing up well to analysis, even at the lower end of the MIPS scale. In the latest issue of Arcati’s “Mainframe Market Bulletin,” pricing analyst Barry Graham argues that the upfront price can be deceptive. “With an entry price of $1 million for almost all users,” he says, “the z10 may seem out of reach to users who have followed a path of purchasing older, second-user systems for the past few years. Yet despite this relatively high minimum entry price, the z10 makes economic sense for most small- to medium-size installations.” Graham goes on to show how substantial reductions in ongoing software and maintenance costs mean that the capital cost of a z10 can be recovered in just over two years, even for a z900 user growing at a modest 10 percent. For sites with more rapid growth plans, of course, the payback time can be considerably shorter.