Having had many years of experience as a sysadmin, I have one word for those concerned with hard drive reliability - temperature. Heat is the enemy of hard dri

Having had many years of experience as a sysadmin, I have one word for those concerned with hard drive reliability - temperature. Heat is the enemy of hard drives (and other system components). Sometimes heat causes a machine room meltdown, where machines start failing left and right. This is not uncommon. What usually happens is this - the company knows enough to put their web server etc. at a controlled colocation facility (which isn't always ideally controlled, but that's another story). But the developers would like a local file server, development source control server etc. There is a spare, windowless room in the office, and without much planning, a machine or two goes in. Then a machine goes in attached to a tape drive, which backs up those machines and the local desktops. Then more machines go in, then more. All these machines generate heat, so a room cooler is bought. But it drops condensation in a cup, and shuts off when the cup is full of water. So someone's job becomes to empty that cup every morning. But then summer comes, and on a particularly hot Sunday afternoon the condensation cup fills, the cooler shuts down, the outside temperature combines with the temperature of a closed, windowless room full of machines for hour after hour. Finally one machine has a component fail. Then another. Then e-mails and phone calls and panic starts flying. I have seen this happen more than once, and have heard about it more. It always starts as an ad-hoc, unplanned, "temporary" solution for "unimportant" machines. But as time goes on, and machines are added, and business dependencies are formed, you have to start supporting a machine room that was never planned as a machine room.