RAID 5
RAID 5
Block-level striping with single distributed parity. Survives any 1 drive failure. The classic choice for 3-4 drive arrays with small drives — but increasingly risky for large drives.
How it works
Data and parity blocks rotate across drives. For each stripe, one drive holds the parity computed from the others. When a drive fails, parity + remaining data reconstructs the missing bytes. A URE during rebuild kills the array — there is no second parity to fall back on.
Formula: (N − 1) × min(drives)
Pros / Cons
Pros
- Survives 1 drive failure
- Highest capacity efficiency of any parity RAID: only 1 drive of overhead
- Fast reads (parallel across all drives)
- Universally supported
- Good fit for small drives (≤8 TB) in 3-4 bay setups
Cons
- A URE during rebuild can kill the array
- Long rebuild times with large drives (20+ hours for 8 TB)
- Write penalty: each write requires read-modify-write of parity
- Not safe with drives ≥ 12 TB on consumer-class drives
- No protection against bit rot
When to use
3-4 drive arrays with drives ≤ 8 TB. Home NAS with low-to-moderate write workload. Cases where capacity efficiency matters more than maximum safety. Always pair with off-site backups.
When NOT to use
Drives ≥ 12 TB — URE probability during rebuild becomes too high. 5+ drive arrays — more URE exposure. ZFS pools — use RAIDZ1 with checksums. Mixed sizes — use SHR-1 or Unraid 1P.
Rebuild math example
4 × 8 TB in RAID 5. After one drive fails: reads (4-1) × 8 = 24 TB from surviving drives. At 70 MB/s rebuild speed, takes ~32 hours. URE probability with NAS-class drives (~10⁻¹⁵/bit) is roughly 17%. A single URE = array loss.