Wednesday 14 May 2008

Western Digital Raptors and Intel On Board RAID Caveats

We seem to be experiencing some hardware related hiccups specific to the 150GB Western Digital Raptors when mirrored on an Intel on board RAID controller.

We have a large number of Seagate RAID 1 mirrors on Intel on board RAID controllers and to date have not experienced this same issue under similar circumstances.

What we have been seeing is the breaking of the mirror if there is a hardware stall or freeze due to either an incorrectly installed/configured hardware driver or a malfunctioning piece of hardware that locks up the system.

So far, the breaking of the mirror seems to be isolated to the Raptors.

In one case, we had a mirror break that also took out the OS.

Keep in mind that we are only installing the Raptors into high end systems that we build in-house for key clients. This is our first run back into Western Digital drives in over five years.

However, any systems that we have put together so far that have displayed the above issues have indeed experienced the RAID 1 mirror being broken and at least needing an array rebuild or in some cases an array reinitialization and now in one case a reload of the OS.

On initial observation, it looks like the Raptors are a little more sensitive to a break in the data stream.

Philip Elder
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.


Anonymous said...

Very interesting, I had similar problems recently with a client running Vista, the client asked that the computer (Intel Bad Ax Motherboard) be upgrade to use 2 150 GB Raptors using the on board raid, the cloning to the Raptors went well computer booted fine, Raid 1, after about 12 hours use client called and said computer was running check disk. Array was degraded. After many repair attempts replaced the Raptors with two 750 GB Seagate Drives in Raid 1and everything is fine. Thanks for the great blog!


Anonymous said...

Tried and stopped using WD enterprise SATA's on an adaptec SATA RAID controller a while back now. I lost the raid array two out of two servers a month into production, it was blamed on the drives internal checks causing a timeout on the RAID controller which eventually marked the drive as bad kicking off the Hotswap, if two went on the RAID5 it was goodbye to the array.

Philip Elder Cluster MVP said...

So far, every build we have done on the drives has had some sort of hiccup or other with the RAID configuration.

We have one more pair of the drives in stock, and we may consider them to be the last depending on the systems we now have in our client's hands performance.

If we end up getting a lot of grief we will ShadowProtect the OS onto a new set of Seagates at no charge ... then firesale the Raptors! ;)

Thanks for the comments folks!