[sard420]

We used to make PODs like this at work for netbackup storage.. started with WD drives ~128 WD drives, at about two years we were at ~8 WD drives still working. At least once a week we had people in there swapping drives. I have yet to see a WD not fail.

[dublinben]

On the other hand, I've got a WD Caviar Black that's been running strong for more than five years. None of my WD drives have ever failed (knock on wood!). I don't run 100+ drives though, so I'm probably just hiding in the statistical margin.

[miahi]

Your Caviar Black is trumped by my two Scorpio Blacks I had to RMA in one year because of strange read/write delays and clicking noise. No actual failure, but it started from one problem/day when I bought them to 3-4/hour when I sent them back.

[dspillett]

Also your thermal environment is going to be quite different, which could be a significant factor. How often do you power-cycle that drive BTW? I've had more drives go bad on power-up (just failing to spin up and move the heads correctly) than through surface errors and such.

[dublinben]

I try not to restart that machine more than once a month.

[dspillett]

That si better than running the drives 24/7 - if one is going to go then that way it is more likely to go alone. If every in an array (anything I even half way care about get s the RAID1 or RAID5 treatment) has been spinning for years then a power cycle can be quite scary. Restarting once month or so increases the chance that if they fail they'll not all fail at the same time.

Having said that, my main little server at home hasn't had a power cycle for somewhat longer than a month as there have been no kernel updates or that like that have required it...