/ Sign-up
Your question

Sata vs Sas

  • Hard Drives
  • SAS
  • SATA
  • Servers
  • Western Digital
  • Storage
Last response: in Storage
August 6, 2011 7:00:46 AM

Hello guru's!

I have a server with 16 WD Black 1TB drives running in raid. These are nice sata drives. However, I lose about six a year. I suspect heat. My warranty replacement is going to run out at some point... I'm wondering what the current thinking is on replacing these with SAS. I didn't use SAS when I first populated the server because of cost. Now I'm wondering if the SAS drives would be less prone to failure.

Any thoughts?

More about : sata sas

a b G Storage
August 7, 2011 10:44:00 AM

I have run a lot SATA and SAS drives in large RAID arrays, its not the interface that makes the difference. but the quality of the drives. Since SAS tends to be enterprise class technology, all SAS drives tend to be made with 24x7 datacenter environment in mind where as the same cannot be said of all SATA drives. Switching to SAS will probably give you better results, but not simply because its SAS instead of SATA. Enterprise quality SATA drives will give you about the same results. WD stopped official MTBF numbers.

If you look at Near-Line SAS, these are essentially enterprise class SATA drives with SAS interface.

The WDs are not unreliable drives. If you suspect heat, I would defintiely correct that first and see if that extends your drive life.
a c 113 G Storage
August 7, 2011 12:16:05 PM

Are the drives really defective or do they drop out of the array? Using 16 drives not designed for RAID use might be the issue. WD RE4 drives should work fine, but 10K or 15K SAS drives are faster.
Related resources
Can't find your answer ? Ask !
a c 415 G Storage
August 7, 2011 3:52:52 PM

^+1 to both of the above posts, particularly GhislainG's. Consumer hard drives don't use a TLER (time-limited error recovery) strategy and this often causes problems for RAID controllers. If the drive is having difficulty reading a sector it keeps retrying over and over again, and this causes the array to give up on the drive and declare it to be dead. With an Enterprise-class drive the drive will simply report an unrecoverable sector, and this will cause the RAID controller to re-write the sector to the drive (assuming that it's a redundant RAID organization that allows the data to be recovered from the other drives). When the sector is re-written the drive remaps it to one of its spares and everything carries on as if nothing happened.
August 8, 2011 3:40:34 AM

Thanks for the replies.

The "guts" of the drive was my real question... I realize the interface is a small issue and know that the SAS drives are typically better hardware. Hence the price increase.

TLER is settable on this series of drives I am using, and I do have it configured properly. That's the not the problem.

I suspect my trouble is heat related, but don't know for certain. I'm using a Norco case. It's not the best for air movement. Not positive this is a heat issue though, as eventually a drive will just drop out of the array and not be able to be read any longer. WD has been good about replacing them so far.

I'll experiment with cleaning up the wiring in the case with hopes that will improve air flow. Maybe that will stop the drives from dropping out.

Technology = fun!