RAID Calculator

Usable space, fault tolerance, and rebuild risk for RAID 0, 1, 5, 6, and 10 — with mixed drive sizes and honest URE math.

Usable capacity
Raw capacity
Storage efficiency
Fault tolerance
URE risk during rebuild

Mixed drive sizes are truncated to the smallest drive on parity and mirror levels. Rebuild risk is the chance of hitting an unrecoverable read error while reading the surviving drives — a worst-case floor, not a guarantee. RAID is not a backup.

What each level trades

RAID combines several drives into one array, trading capacity for speed, redundancy, or both. The five common levels sit at different points on that trade-off:

LevelUsable (n drives, size s)Survives
RAID 0n·s (striped)nothing — any failure loses all
RAID 1s (mirrored)up to n−1 drives
RAID 5(n−1)·s1 drive
RAID 6(n−2)·s2 drives
RAID 10(n/2)·s1 per mirror pair

Mix drive sizes and the parity and mirror levels can only use the smallest drive's worth from each — a 4 TB drive in an array of 2 TB drives contributes just 2 TB, and the rest is wasted. RAID 0 is the exception; it simply sums.

TB versus TiB

Drive makers sell terabytes as 10¹² bytes, but operating systems usually show tebibytes (2⁴⁰ bytes). That is why a "12 TB" array reports as 10.9 TiB — no space is missing, the two units just differ by about 9%. This calculator shows both.

The rebuild-risk argument, honestly

When a drive fails in a parity array, rebuilding onto a replacement means reading every remaining drive in full. Drives have a specified unrecoverable read error (URE) rate — often one bit in 10¹⁴ for consumer disks. Reading a 12 TB RAID 5 rebuild is about 9.6×10¹³ bits, giving a worst-case URE probability of 1 − e^(−0.96) ≈ 62%. That alarming figure is behind every "RAID 5 is dead" article.

The caveats matter, though. The URE spec is a conservative floor; real drives routinely beat it by an order of magnitude. Many UREs land in unused space or are recoverable at the filesystem layer. And enterprise drives are rated at 10¹⁵ or better, which drops the same rebuild risk below 10%. What survives all the caveats is the practical conclusion: for large arrays, prefer RAID 6 (its second parity block absorbs a URE during a single-drive rebuild), keep arrays from growing without bound, and remember that RAID is not a backup — it protects against drive failure, not deletion, corruption, or fire.

Related tools: the data transfer time calculator estimates how long a rebuild actually takes, and the subnet calculator is the other everyday homelab tool.