Following yesterday's story about how users are keeping systems for longer, we take a look at the opposite point of view.

Mike Sink used to try to get six to seven years out of his IT equipment, but trying to manage older machines, replace worn-out units and hand equipment down the food chain as systems aged was just too much work, he says. Now everything from servers to PCs and data centre equipment at Kichler Lighting is on a 36-month replacement cycle.

Sink, director of infrastructure, cites several reasons for switching back to a three-year replacement cycle. "Expertise with equipment tends to dwindle as new equipment comes out," he says, and older systems don't handle newer applications or operating systems very well.

He also prefers to be proactive and trade IT equipment regularly rather than risk failures -- even if the risk is relatively small. For example, he replaces drive arrays every three years. While Sink hasn't had an array go down -- drives are hot-swappable and fault-tolerant -- he worries that a double fault will take an array out of service. "I'm not saying that the drives aren't good after three years, just that failure is more likely," he says.

Could Sink save money by extending the useful life of machines? Possibly. But he doesn't think it's worth the bother. "When the warranty is up, we bring in a new machine. It eliminates a lot of issues for us," he says.