On 03/12/2013 01:06 PM, Lonnie Olson wrote: > Important thing to consider is Murphy's law. The minimum number of > disk failures to destroy your whole array. In RAID6, it's 3 disks. > In RAID10, it's 2 disks. Obviously this would have to be a matched > pair, but s#!* happens.
RAID 6 stores the parity on 2 disks. So really you can only lose any two disks in a RAID 6 volume before you have catastrophic data loss. So laying aside the rebuild times for RAID 6 which can be very long indeed, if you have a bunch of RAID-6 volumes made up of 4 disks total, then that's the same redundancy as RAID-10, except that you can lose any two out of four disks in a RAID-6 volume, whereas with RAID-10, if you happen to lose two disks that are part of a RAID-1 volume, you are screwed. /* PLUG: http://plug.org, #utah on irc.freenode.net Unsubscribe: http://plug.org/mailman/options/plug Don't fear the penguin. */
