Alan Bourassa, CIO of EmpireCLS, a limousine service in Secaucus, N.J., is a man on a tight schedule. He has tied his company's 64 server blades to a three-year replacement cycle, a fast-paced strategy that he feels has become essential in this era of rapid technological change and unprecedented levels of business pressure.
"A lot of folks try to say they'll keep their blades for five, seven years, but it just doesn't happen because of the way the industry and technology is changing so fast," says Bourassa, who until recently managed a replacement cycle exceeding five years.
Bourassa, like other IT leaders, has compressed his server replacement cycle to the shortest practical length to obtain the flexibility necessary to stay on top of such important trends as virtualization, evolving server and software technologies, expanding storage demands and growing business mandates.
Darin Stahl, a server industry analyst at Info-Tech Research Group in London, Ontario, observes that server replacement cycles these days generally run toward three years, depending on the machine's role and its owner's needs. "It's a scheduled management of assets that lines up with depreciation but also with stability and the useful life of the boxes," he says.
Richard McCormack, senior vice president of Fujitsu America's server and solutions business in Sunnyvale, Calif., says that virtualization -- with its requirement for fewer yet more powerful machines -- is the primary force accelerating the pace at which data centers of all sizes are updating their server inventories.
"Not only has virtualization dramatically increased the level of application/server consolidation, it has also radically changed customer deployment models," McCormack says. Customers are now more likely to speed up server replacement cycles to take advantage of faster processors and other technological improvements, he explains.
Bourassa's top concern at the moment is making sure his Hewlett-Packard servers have enough capacity to support effective virtualization. "I think that folks that are going the virtualization route will be on a faster [server replacement] cycle for the next few years because the capacity of the servers has tripled to quadrupled over the past year," he says.
Faster and cheaper
Paradoxically, shorter server replacement cycles are actually helping some enterprises trim their server budgets. Stahl notes that server replenishments spurred by virtualization typically save enterprises money by allowing them to purchase just a few virtualized machines to replace a larger number of older boxes.
"Right now, the practice in most production shops is getting a 7-to-1 [consolidation] ratio," he says. "So instead of refreshing seven servers, they're refreshing one and getting rid of six physical servers."
On the other hand, a large virtualized server typically costs more to acquire and operate than any of the discrete physical servers it replaces. Therefore, it's important not to fall into the trap of making a direct 6-to-1 cost-savings calculation.
Tighter replacement cycles in a rapidly consolidating environment also make server allocation a far more intricate and increasingly fine-tuned task for many managers. Chris Nowak, CTO at Anthony Marano Co., a fresh produce distributor in Chicago, feels that the best way to keep pace with rapidly growing IT and business demands is to keep his shop in perpetual motion, which entails maintaining a schedule in which he's always evaluating or replacing at least some servers.