SSD Caching and Intel RST

[fletchermac]

I built a new PC in January for video editing. Specs:

MB: AS Rock LGA1155 DDR3 SATA3 USB3.0 Quad CrossFireX and Quad SLI A GbE ATX Motherboard Z77 EXTREME4

Intel i5

RAM: Corsair Vengeance 4x8g

I loaded my OS onto a 64g SSD and installed a 1TB spindle to use as a storage drive. Well, 64g was not enough space for much more than my OS, and that was a problem, so I reinstalled my OS on a raid0 array, using Intel RST with my SSD as a cache drive.

I have had nothing but problems ever since. Intel RST reported a failed volume (spindle), so I replaced about 3 months ago. Today I got the same thing....RST says that my 1TB drive has failed.

So either Intel RST kills hard drives, or I set something up incorrectly. Not sure what that would be, but I was pushing the limits of my knowledge setting up a cache drive, so it's possible that I missed something.

Anyone have any ideas before I go out and buy another new HD and reinstall my OS....again?

 

[Cerb]

If HDDs are actually failing left and right, the PSU would be my gut feeling.

Are the drives failing, or the RAID array? An inconsistent array state would cause data loss, but not necessarily mean a drive problem. The easiest thing to do, when you go buy a new drive, is to get a single drive that's big enough, and not use RAID 0. If you need RAID 0 for the bandwidth, make the RAID 0 volume a data-only volume, separate from your OS/apps.

 

[LoveMachine]

RST is generally pretty fool-proof (Please note, I'm not calling you a fool), and caching RAID doesn't really do much physically to the HDD, only storing blocks of data on the SDD that's pulled from the HDD as it would be without caching. How are the HDDs failing? In other words, after RST reports the failed volume, is the drive still usable? Still able to install Win without RST?

 

[BrightCandle]

Raid 0 doubles your exposure to failures. So while its quicker its a lot more unsafe than a single drive. One of the extra problems is that typically you buy two drives at the same time and they often come from the same batch and are relatively near to each other in the production. Its not uncommon for them both to have similar failure times, its one of the reasons why RAID 5 and 6 aren't as effective in practice as the basic statistics of completely random failures would suggest.

Obviously check the drive has actually failed with it on its own but there is every possibility you have just been hit by a well known issue with matched drives.