2

For 3 years we had an LSI SAN with 48, 300GB Segate Cheetah 15K.5 (Model ST3300655FC) 3.5-inch drives. There are about 7 drives failed total. There bulk failed recently. Six drives since May 2010.

That's at a rate of 0.02 (drives failed)/(month)/(drives in array) for the last 6 month period.

There is an older SAN from HP running in the same room, I the drives are probably 15K 36 GB. Those never failed.

Is it common that 300GB 15K RPM drives start failing at this rate after 3 years?

2
  • 1
    You may have other problems like resonance. Are the failed drives from the same bays?
    – Rob Olmos
    Nov 20, 2010 at 5:19
  • No - different bays all over the 3 shelf array. Nov 20, 2010 at 9:36

3 Answers 3

3

That doesn't sound excessively high to me but as you're concerned here are a few things to check:

  • Physical mounting of the drives, which may well allow the resonance mentioned by Rob Olmos. Check all screws are still tight, in both the drives, their carriers, back plane, etc.
  • Power supply/supplies, which may have drifted off spec and could be placing additional electrical strain on the drives and other electronic components.
  • External vibration affecting the drives and/or entire unit (the biggest problem in my server room).
1
  • +1 I'll try to tighten (and/or put in) the skews that fasten the enclosures to the rack. Thanks! Nov 20, 2010 at 9:42
4

No, this is not a common problem, but one I've seen every now and again. It's likely that the Cheetah 15ks you're running are all from the same "batch" in manufacturing and as such, all have the same faults. This is why you're seeing many of them fail in such a short amount of time.

I'd recommend contacting your SAN vendor and insisting that they make this right.

3

Well there is the MTBF inverted bell-curve failure issue that you'd expect to see, so after three years if you start seeing one or two disks failing then I'd be expecting a few more soon after as you've seen. It then usually settles down into a series of small peaks and troughs of reliability - best to just ensure you've got some spares close to hand. As for why the HP isn't failing, well the physics are the same, I'd imagine their behaviour is just influenced by their usage patterns but when they do start failing it's likely to be in the same manner.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .