RAIDZ2 Calculator

Q: What's the minimum number of disks for RAIDZ2?

4 disks, matching RAID 6's minimum, though at 4 disks a ZFS mirror is also worth considering given the 50% overhead.

Q: Does RAIDZ2 protect against silent data corruption?

Yes — ZFS checksums every block and can detect and repair silent corruption using parity data, which traditional RAID controllers generally cannot do on their own.

ZFS RAIDZ2 is double-parity RAID built into ZFS — the recommended default for most ZFS pools of 5 or more disks, tolerating two simultaneous disk failures.

Number of Disks

Disk Size (TB)

RAID Level

The RAIDZ2 formula

Usable Capacity = (N − 2) × Disk Size

Same N-minus-2 math as hardware RAID 6, built directly into ZFS's copy-on-write filesystem instead of a separate RAID controller layer.

Worked example

6 disks × 4 TB = 16 TB usable out of 24 TB raw, 66.7% storage efficiency, tolerates 2 disk failures. On the same 6 disks, RAIDZ1 would give 20 TB (83.3% efficiency) but only survive 1 failure.

Why RAIDZ2 is the recommended default

ZFS resilvers (its term for rebuilds) on large modern drives can take many hours, during which a RAIDZ1 vdev runs with zero redundancy — the same rebuild-window risk that makes hardware RAID 5 risky at scale. RAIDZ2 survives a second failure, or an unrecoverable read error, during that exact window. Combined with ZFS's copy-on-write design (no write-hole risk) and checksummed data (catches silent corruption that RAID controllers can't), RAIDZ2 is widely considered the safest practical default for general-purpose ZFS pools.

Keep vdev width in mind

ZFS documentation generally recommends keeping a single RAIDZ2 vdev under roughly 10-12 disks — beyond that, build a second vdev in the same pool rather than one very wide vdev, to keep resilver times reasonable and avoid IOPS bottlenecks.

What's the minimum number of disks for RAIDZ2?

4 disks, matching RAID 6's minimum — though at 4 disks the 50% overhead makes a ZFS mirror worth considering too.

Does RAIDZ2 protect against silent data corruption?

Yes, and this is a real advantage over hardware RAID 6 — ZFS checksums every block and can detect and repair silent corruption (bit rot) using the parity data, something traditional RAID controllers generally can't do on their own.

Need a different RAID level? Use the full RAID Calculator covering RAID 0, 1, 5, 6, 10, 50, 60 and ZFS RAIDZ1/2/3, or read the RAID levels comparison guide.