Can I mix disk sizes in a ZFS RAIDZ vdev?

Technically yes, but ZFS treats every disk in the vdev as the size of the smallest member, so a larger disk just wastes its extra capacity. Use identically-sized disks per vdev.

RAID Capacity Calculator

Q: Should I use RAIDZ2 or RAIDZ1 for a home NAS?

RAIDZ2 is the safer default once you have 5+ disks, for the same reason hardware RAID 6 is preferred over RAID 5: rebuild windows on large modern drives carry real risk of a second failure, and RAIDZ2 survives that.

Enter the number of disks, disk size, and RAID level to get usable capacity, storage efficiency, and how many disks can fail before data loss.

Number of Disks

Disk Size (TB)

RAID Level

Source: standard RAID parity math (RAID 0/1/5/6/10/50/60) and OpenZFS documentation for RAIDZ1/2/3. See the full methodology page. Reviewed: July 2026.

How RAID levels affect usable capacity

Every RAID level trades raw disk capacity for either speed (RAID 0), redundancy (RAID 1, 5, 6, 10), or a mix of both (RAID 50, 60). This calculator applies the standard capacity formula for each level so you can size a volume before you buy the drives. New to RAID? Start with What Is RAID? or the full RAID levels reference.

Looking for one specific level? RAID 0 · RAID 1 · RAID 5 · RAID 6 · RAID 10 · RAID 50 · RAID 60 · RAIDZ1 · RAIDZ2 · RAIDZ3

Choosing between RAID 5, RAID 6 and RAID 10

RAID 5 tolerates one disk failure and has the best space efficiency of the parity levels, but rebuild times on large modern drives (8TB+) carry real risk of a second failure during rebuild — many admins now default to RAID 6 for arrays with more than 6-8 large disks. RAID 10 gives the fastest rebuilds and best random I/O at the cost of 50% usable capacity, and is the standard choice for databases and VM datastores.

What's the minimum number of disks for RAID 6?

RAID 6 requires a minimum of 4 disks, and tolerates any 2 simultaneous disk failures.

Is RAID a replacement for backups?

No. RAID protects against disk failure, not against accidental deletion, ransomware, or site loss. It should be combined with an offsite/offline backup strategy.

How is RAID 50/60 usable capacity calculated?

RAID 50 and 60 stripe across multiple RAID 5 or RAID 6 sub-arrays. This calculator assumes disks are split into 2 equal sub-arrays (a common default) — for other sub-array counts, treat this as an approximation.

ZFS RAIDZ1/2/3 — similar math, different mechanics

RAIDZ1, RAIDZ2, and RAIDZ3 are ZFS's own parity RAID implementations, and the usable-capacity math tracks RAID 5/6 closely: RAIDZ1 loses one disk's worth of capacity to parity, RAIDZ2 loses two, RAIDZ3 loses three — but tolerates three simultaneous disk failures, which no traditional hardware RAID level offers. The real differences are underneath the math:

No write hole. Traditional RAID 5/6 has a known "write hole" risk — a power loss mid-write can leave data and parity out of sync. ZFS is copy-on-write: it never overwrites data in place, so a RAIDZ vdev can't develop this specific inconsistency the way hardware RAID controllers can.

Recommended vdev width matters more than the math suggests. ZFS documentation and most experienced admins recommend keeping a single RAIDZ vdev in a specific disk-count "sweet spot" — commonly cited as up to around 8-12 disks per vdev depending on RAIDZ level and disk size, not the 16-20+ some hardware RAID 6 arrays tolerate. Beyond that, resilver times and IOPS-per-vdev limitations degrade performance more than the raw capacity math implies; build multiple RAIDZ vdevs in one pool instead of one very wide vdev.

You historically couldn't add a single disk to an existing RAIDZ vdev — expanding meant adding a whole new vdev to the pool, not growing the existing one disk-at-a-time like some hardware RAID controllers allow. OpenZFS 2.3 introduced RAIDZ expansion (adding one disk at a time to an existing vdev), but plenty of production systems still run older OpenZFS versions without it — check your specific version before assuming this is available.

dRAID is a separate, newer ZFS vdev type (distributed spare RAID) built for very large pools with fast rebuilds using distributed spare space rather than a dedicated spare disk — out of scope for this calculator's simple N-minus-parity math, but worth knowing about if you're planning a large all-flash or many-disk pool.

Should I use RAIDZ2 or RAIDZ1 for a home NAS?

RAIDZ2 is the safer default once you have 5+ disks — same reasoning as hardware RAID 6 over RAID 5: rebuild/resilver windows on large modern drives carry real risk of a second failure, and RAIDZ2 survives that. RAIDZ1 is more defensible on smaller arrays (3-4 disks) where resilver time is shorter.

Can I mix disk sizes in a RAIDZ vdev?

Technically yes, but ZFS treats every disk in the vdev as the size of the smallest member — a mismatched disk larger than the rest just wastes its extra capacity. Use identically-sized disks per vdev to avoid this.