we have several dell servers with raid 5. during the time we have had them, 3 hard drives have failed. in two of those cases, it seemed that the one drive going down knocked off another drive, crashing the server. one of the drives was able to be rebuilt afterwards both times.
dell says that this happens sometimes. if this is going to happen 67% of the time a drive fails, then raid for fault tolerance is worthless for us. dell says the odds are about 10% of one drive knocking another off, is that industry accepted standard?
one possibility would be switching to raid 1+0 (if i have my terms right). i'd set up two controllers, with three drives each. stripe three drives across one controller, and mirror them on the three on the other controller. my hope would be that a drive going down couldnt knock off a drive on a separate controller.
has anyone had any experiences like this? any suggestions for how to handle this?