Having just watched engineers try and trouble shoot a RAID 5 set with multiple drive failures, built on Microsoft software RAID, I'd like to chuck my hat in the arena on this subject...
Don't make a software RAID 5 set!
The issue with software RAID 5 sets (especially with Windows OS) is that there is no straight forward way of monitoring the drives within the array, and just as important (as the engineers I was with discovered) no easy way of identifying the drives to establish which are the missing devices!?
Modern hardware RAID controllers are ridiculously cheap and perform a vital role in maintaining the resiliency of your data. These cards also offer a vital source of drive behaviour through active monitoring and at the very least can email you when drives start to misbehave.
Be cognizant of read errors as they are a serious issue with RAID. If the disk fails, the RAID will have to read all the disks that remain in the RAID group to rebuild. A read error during the rebuild will cause data loss since we need to read all other disks during rebuild.
ReplyDeletehttp://www.zdnet.com/article/why-raid-5-stops-working-in-2009/
SATA disks experience read failure once every 12,5 TB of read operations, that means if the data on the surviving disks totals 12.5TB, the rebuild is certain to fail.
http://www.lucidti.com/zfs-checksums-add-reliability-to-nas-storage