RAID 6

RAID 6 is a way of storing data across several disks with two independent sets of parity, so the array keeps working and loses no data if any two disks fail.

Like RAID 5, RAID 6 spreads data in stripes across its disks. Each stripe carries two calculated values, usually called P and Q, instead of one parity block. With two, the array can work out any two missing blocks in a stripe. The cost is capacity: storing N disks’ worth of data takes N + 2 disks.

How is RAID 6 different from RAID 5?

RAID 5 survives one failed disk and RAID 6 survives two. In normal use and with one disk missing, RAID 6 performs close to RAID 5. With two disks missing it still serves data, but it is very slow, because every read of a missing block becomes a calculation from the disks that remain.

What happens when a RAID 6 array loses a third disk?

The volume stops, because some stripes can no longer be calculated. Two missing disks leave no redundancy, so the array is already a degraded array with nothing to spare, and a third failed or unreadable disk takes away blocks that can’t be worked out from the rest. See RAID data recovery.

Common misconceptions

Does RAID 6 mean I don’t need a backup?

No. RAID 6 protects against disks failing, not against deletion, overwriting or a damaged file system. A deletion or an overwrite is a change like any other: it is written to every disk it touches, parity included, so the array can’t undo it.

Sources

Back to the glossary