ZFS RAIDZ3

Triple-parity ZFS. Maximum safety for very large pools where any failure during rebuild must not cause data loss.

Min. drives
5
Usable capacity
(N-3) × smallest
Fault tolerance
3
Performance
Fast reads, slower writes

How it works

RAIDZ2 plus a third independent parity. Tolerates 3 simultaneous drive failures. Designed for wide vdevs (10+ drives) where the long resilver windows make 1-2 parity insufficient. Same checksumming and used-block resilver as RAIDZ1/2.

Formula: (N − 3) × min(drives)

ZFS RAIDZ3 — D = data, P = parity, Q = second parityD1D1PQD8D2PQD5D9D3QD2D6PD4D3D6PQD5D4PQD7
Layout diagram

Pros / Cons

Pros

  • Tolerates 3 simultaneous failures
  • Strongest URE protection of any RAID
  • Full ZFS integrity stack
  • Designed for large vdevs (10+ drives)
  • Free, open source

Cons

  • Loses 3 drives of capacity
  • Lower write performance
  • Overkill below 8 drives
  • Long resilver times for very wide vdevs
  • Same vdev-expansion awkwardness

When to use

Pools with 8+ drives where uptime matters more than capacity. Mission-critical archives. Cold storage at scale.

When NOT to use

Small pools (≤ 8 drives) — RAIDZ2 is fine unless you really need triple parity.

Rebuild math example

12 × 8 TB RAIDZ3 at 80% fill: resilver reads (12-3) × 8 × 0.8 = ~58 TB at 70 MB/s ≈ 76 hours. Triple parity tolerates even multiple UREs during the long resilver.

Check rebuild risk →

Try it now

Related

ZFS RAIDZ2 →The ZFS sweet spot. Double parity, checksums, snapshots, used-block resilver. Strongly rec... RAID 6 →Block-level striping with double distributed parity (P + Q). Survives any 2 simultaneous d... Unraid (2 parity) →Double-parity Unraid. Drive isolation + 2-failure tolerance. The Unraid choice for 5+ driv... Browse all RAID levels →RAID Levels
Bottom line: RAIDZ3 buys peace of mind for wide vdevs. Below 8 drives, RAIDZ2 is the better balance of capacity and safety.