I had a quick question about RAID. I understand that RAID uses one or more parity disks to repair failed disks, but I was wondering what they mean by "declustering" the disks? And also, if someone could explain the concept of "scrubbing" when it comes to disks and such (in a way that a beginner could understand) that would really be helpful :)
Declustering uses multiple RAID controllers in a way that a failed disk can be recovered by work done by more than one controller at a time. Otherwise, if you lost a single disk in a RAID 5 or RAID 6 array, the rebuild would take about as long as it takes to read all the data from an array drive, which can be very long with modern multi-TB drives.
Scrubbing resolves the problem of hidden data errors. A scrubber continuously (or on a schedule), at a slow rate and a low priority, reads every byte of data from every drive. If any data cannot be read, the data is recovered from the parity drive(s) and re-written. This ensures that errors do not accumulate and it ensures that drive failures are detected as early as possible. This increases the chances that a rebuild after a failed drive will succeed.
Every decent RAID controller has some kind of scrubbing capability. The most basic would be to scrub every 24 hours at low priority. Ideally, hot spare drives would be scrubbed as well. In contract, declustering is in very limited use right now.
RAID by itself does not necessarily use one or more disk for parity -- it all depends on the RAID level. For example, RAID 0 has no parity at all. Instead it is built for pure performance.
For more information about RAID levels you can visit http://en.wikipedia.org/wiki/Standard_RAID_levels
As far as "declustering" and "scrubbing", can you provide either a link to this discussion, or addtional details?