Dell says: "Firstly, run data centres at 25 degrees Celsius rather than the usual 16-18 degrees. Significant cooling costs can be obtained and the data centre IT kit runs perfectly well."

Hmm prior studies seem to suggest you knock roughly one month off the lifetime of a server for every degree ceclius increase. Of course, a kit maker would probably see this as a benefit :-).

but Dell is a hardware supplier so of course running servers warmer will be beneficial - to Dell. As we'll need to buy new hardware to replace the dead stuff.

I'd like to see the hardware failure rates at different temperatures *and* tally that with the cost of replacement hardware versus cooling cost savings. If the replacement hardware cost is less than the savings made in AC costs, then it clearly is beneficial.

I would suggest the idea of upsizing hard disks to take advantage of the fact that these disks are becoming cheaper and larger in data capacity. Same too with the RAM. Then you are gaining more mileage from your existing servers.

We have tried several times to make better use of older hardware by increasing RAM and Hardrive capacity and introducing virtualisation for systems that don't require a large amount of fast cpu but each time we've been stymied by the hardware companies not supporting the older models because they've moved to a different architecture/model. It's all very well for a hardware manufacturer to make the claims that Dell do but it's unrealistic for most companies - if we can get support for our older kit, it costs more than it's worth once the kit becomes EOL which seems to come sooner and sooner these days.