Today we have a story about one school taking a stronger stance on conserving energy by keeping the servers up and running during the hot days of summer, but slowing the servers down. Any eco benefit is apparently simply a bonus, as the technique was apparently created for an emergency when servers might be heating up due to some cooling failure. Here are some fascinating particulars.
1. Systems that benefited the most from the slowdown during a data center cooling event were AMD systems, not Intel.
2. The servers in question were running Linux, not something else.
3. In two separate cooling events, the slowdown concept, created by a very talented and clearly brilliant admin, worked well enough to keep the servers running despite a massive problem taking place behind the scenes. I know I wish my own personal Web hosts had such a concept going for them!
Now here is the really awesome part. The program, developed from a common sense concept, is going on the road and is now available to everyone. Yes, it looks like those using AMD-based clusters can now benefit from this innovation. It’s a neat idea and one that I think is totally worth looking into. Is it worth it to you? Well, it depends. If your specs match this guy’s and you believe that having a throttle back option in a crisis to deal with a lack of cooling is a big of a enough deal, then buying this guy’s software is a no-brainer.
[awsbullet:Thomas L. Friedman energy]