Parity is the clever maths that lets RAID 5 and RAID 6 rebuild a failed drive without keeping a whole second copy. Here's how it works — then see the capacity and rebuild trade-offs in the RAID calculator.
Parity in one sentence
Parity is a checksum, computed across the data blocks in a stripe, that lets you reconstruct any one missing block from the others. RAID 5 stores one parity block per stripe; RAID 6 stores two independent ones. The classic single-parity calculation is an XOR: if you know all the data blocks but one, plus the parity, you can recover the missing block.
Because parity is one block per stripe rather than a full copy, it is far more space-efficient than mirroring — RAID 5 keeps (n−1)/n of your raw capacity usable. That efficiency is why parity RAID dominates capacity storage.
Single vs dual parity
RAID 5's single parity survives one drive failure. RAID 6 adds a second, mathematically independent parity (using Reed-Solomon / Galois-field maths, not just XOR), so it survives any two failures — and, crucially, a read error during a single-drive rebuild. That rebuild safety is the main reason RAID 6 replaced RAID 5 on large drives (see RAID 5 vs RAID 6).
Triple parity (RAIDZ3, RAID-TEC) extends the idea to three checksums for very wide arrays of huge drives.
The cost: the write penalty
Parity isn't free to maintain. Every write must update the parity too, which means extra back-end I/O — a ×4 write penalty for RAID 5 and ×6 for RAID 6. That's why mirror-based RAID 10 (×2) is preferred for write-heavy databases. See the write penalty explained.
The calculator shows the write penalty and resulting IOPS for any level so you can weigh space efficiency against write performance.