RAID failures

Multiple Disk Failures in RAIDs

Fault-tolerance is an ability of RAID not to lose data if a failure occurs. If RAID can survive a disk failure with a probability of, say 78%, such RAID is not a fault-tolerant array. The survival probability must be 100% for an array to be considered fault-tolerant. However, in practice it is useful to know probabilities of failures.

Let's consider various RAID array layouts in terms of fault-tolerance. RAID 0 is not a fault-tolerant array, RAID 1, RAID5 and RAID10/50/60 are fault-tolerant and can survive a single disk failure, while RAID 6 can survive a failure of two member disks.

Simultaneous multiple disk failure is an event, when a failure of the second and the subsequent disks happens before the rebuild (caused by a failure of the first disk) completes. Since rebuild of modern RAIDs can take hours, and often days, the term "simultaneous disk failure" becomes a vague concept.