Rebuild Time & URE Risk Calculator

When a drive fails in a parity RAID, the array rebuilds onto a replacement. Two questions matter: how long will it take? and how likely is the rebuild to fail? This tool answers both, based on standard RAID math and manufacturer-published URE rates.

Your configuration
~1 read error per 125 TB read. Common for NAS-grade drives.
TB
×
MB/s
30-70 typical under load, 100-200 fully idle. Conservative default.
%
Only ZFS resilvers used blocks. Block-level RAID (RAID 5/6, Unraid, SHR) always reads the full drive during rebuild.
Results
Verdict
How this is calculated

Rebuild time = drive size × fill / rebuild speed. Real-world speed depends on controller, ongoing I/O load, and drive performance. Idle arrays reach 100-200 MB/s; production arrays often drop to 30-80 MB/s. ZFS only resilvers used blocks.

Bits read during rebuild: RAID 5 reads (N-1) × drive size. RAID 6 reads (N-2) × drive size when one drive fails. RAID 10 / RAID 1 only reads the surviving mirror partner.

URE probability: P(≥1 URE during rebuild) ≈ 1 − exp(−r × B), where r is the URE rate per bit (10⁻¹⁴ desktop, 10⁻¹⁵ NAS, 10⁻¹⁶ enterprise) and B is bits read. RAID 5 / RAIDZ1: single URE = array loss. RAID 6 / RAIDZ2: tolerates one URE.

Sources: drive manufacturer datasheets (Seagate Ironwolf, WD Red, Toshiba N300), Backblaze Drive Stats, our methodology.

Takeaway: With drives ≥ 8-12 TB, RAID 5 / RAIDZ1 carries real rebuild risk on consumer drives. Use RAID 6, RAIDZ2, or Unraid with double parity. Try the main calculator →