Temp does not matter as much as has been thought. See: http://labs.google.com/papers/disk_failures.html http://labs.google.com/papers/disk_failures.html IIRC

Temp does not matter as much as has been thought. See: http://labs.google.com/papers/disk_failures.html http://labs.google.com/papers/disk_failures.html IIRC, up to about 40 C, failure rates are pretty flat as a function of temperature. The paper has some nice graphs. There is one other large-scale study in circulation that showed the same thing, but I don't recall the source.