Tool 03 · Storage

RAID & ZFS capacity calculator

Usable space, storage efficiency and fault tolerance for classic RAID and ZFS layouts, including striped vdevs, plus the number most calculators won't show you: the probability of hitting an unrecoverable read error during a rebuild.

Array

TB

Capacity

usable

Efficiency
Fault tolerance
Rebuild URE risk
Data () Redundancy ()

All layouts with these drives

LayoutUsableEfficiencyToleranceRebuild URE risk
Copied

How it's calculated

Usable capacity per group: stripe keeps all w drives, a mirror keeps 1 of w, single parity keeps w−1, double parity w−2, triple parity w−3. Groups stripe together, so totals multiply. A mirror layout with multiple groups is exactly RAID 10, and multiple RAIDZ groups are striped vdevs in a pool. Capacities are decimal TB as sold by drive vendors; the TiB figure is what your OS will report. Real-world ZFS pools lose a further few percent to metadata, padding and the default slop reservation.

Rebuild URE risk is the probability of hitting at least one unrecoverable read error while reading the data needed to rebuild one failed drive: P = 1 − (1 − 1/URE)^bits, where a mirror rebuild reads its surviving copy and a parity rebuild reads the whole remaining group. For double and triple parity the figure shown is the risk during the first rebuild while redundancy still covers a URE. That is exactly why RAID 5 with large desktop drives is a gamble and RAID 6/Z2 exists. Tolerance is per group: losing more drives than the group's parity kills the whole pool, no matter how many other groups are healthy.

This is a planning model: it assumes independent failures and datasheet URE rates, which are conservative for healthy drives and optimistic for a batch with a defect. Pair wide, large-drive groups with a tested backup. RAID is availability, not backup.