2011-11-29 10:48 PM
I have a doubt regarding RAID5 reorganization. Suppose we have a RAID5 group consisting of 6 disks in a storage system. Suppose, one disk (call it d1) is unreachable i.e the disk component does not fail but some component (for example, some network cable) to reach that disk has failed. Although the data in d1 is valid, the data in d1 is inaccessible. In this scenario, will the reorganization will start or some field repair-person will try to repair/replace that component? Replacing a component seems to take less time than reorganization because reorganization is very costly with respect to time . Also, data in d1 is valid (so just repairing that component will make everything fine).
2011-11-30 07:33 AM
NetApp runs RAID_DP and RAID4 (not raid 5) but if a disk fails it automatically rebuilds without user intervention...the user only needs to swap the dead drive. NetApp also pre-fails a disk with predictive failure and does a copy of the drive to a spare before the drive has a hard failure (rapid raid recovery is faster than a full rebuild).