Penton Systems / Tools / RAID
Free tool · storageRAID rebuild risk
Everyone knows RAID 5 is risky with big drives. This shows how risky, for your drive count, size and error rate: how long the rebuild takes, how the chance of losing the array grows while it runs, and how every other level compares on the same drives.
Calculator
This calculator needs JavaScript. The model is described below.
How it works, and how far to trust it
- 01
Rebuild time
Data to rewrite divided by effective speed: drive size × the share of the drive that holds data, over sustained speed × the rebuild priority your controller allows.
- 02
A second failure
With a constant failure rate, the chance another drive fails during the rebuild is 1 − exp(−AFR × hours / 8,760). RAID 6 and triple parity lose data only if two or three more fail in that window, a binomial tail.
- 03
Read errors (URE)
RAID 5 reads every surviving drive in full, so a read error becomes likely on large arrays: P = 1 − exp(−bits read / URE specification). Mirrors and RAID 10 read only the failed drive’s partner.
- 04
What it cannot know
Real drives fail in correlated batches, age together and are stressed by rebuilds, so real risk is higher than these independent-failure numbers. Use this to compare designs, not to promise an outcome. RAID is availability, not backup.
Size the array, the spares and the backup together
We will design the array, recommend the spares and plan the backup, and tell you plainly where the risk sits.