RAID 5 vs RAID 6: which parity scheme should your array use?

The parity array trade-off that gets more important as drive capacities grow.

RAID 5 uses one parity drive and tolerates one drive failure. RAID 6 uses two parity drives and tolerates two. In a 4-drive 8 TB array, RAID 5 gives you 24 TB usable and survives one failure, RAID 6 gives you 16 TB usable and survives two. The choice used to be mostly about capacity efficiency. With drives over 10 TB becoming mainstream, it's increasingly about rebuild safety.

Try this comparison with our tools

Same 8 × 8 TB drives, each scheme

An 8-bay array of 8 TB drives under each scheme. RAID 5 gives you one more drive of usable space (56 TB vs 48 TB), but RAID 6's second parity drive is what survives a failure during the multi-day rebuild that large drives now require.

RAID 5

Usable capacity

56 TB

of 64 TB raw

Storage efficiency — 88%

Fault tolerance

Survives 1 drive failure

RAID 6

Usable capacity

48 TB

of 64 TB raw

Storage efficiency — 75%

Fault tolerance

Survives 2 simultaneous drive failures

Capacity and fault tolerance only — rebuild risk and write performance also depend on drive size and controller.

Size your own array? Open the RAID Calculator

Option A
RAID 5
Wins 3 of 11 compared specs
Option B
RAID 6
Wins 3 of 11 compared specs

Side-by-side specs

SpecRAID 5RAID 6
Parity drives12
Drives tolerated lost12 (better on this spec)
Usable capacity (4 × 8 TB)24 TB (better on this spec)16 TB
Usable capacity (8 × 8 TB)56 TB (better on this spec)48 TB
Efficiency (N drives)(N-1)/N(N-2)/N
Min drives required34
Write overhead~1.25× baseline (better on this spec)~1.5-1.75× baseline
Read performanceFastFast
URE risk during rebuildSignificant at 10TB+Survivable (better on this spec)
Safe for 10TB+ drivesRiskyYes (better on this spec)
Typical use3-4 bay NAS5+ bay NAS / servers

How they differ

The key URE rate (unrecoverable read error rate) is standard. Consumer hard drives spec around 1 URE per 10^14 bits, roughly one read error per 12.5 TB read. On a RAID 5 rebuild of a failed 20 TB drive, the array must successfully read every bit of every other drive to reconstruct the missing data. With six 20 TB drives left, that's 100 TB to read, and the probability of hitting at least one URE during the rebuild is non-trivial (often cited at 25-50% for consumer drive arrays). A URE during RAID 5 rebuild means total data loss. RAID 6 survives a URE during rebuild because it still has a second parity block to reconstruct from. You only lose data if a second drive fails completely while rebuilding. Write performance is the other trade: RAID 6's two parity overhead means writes are 15-25% slower than RAID 5 on the same hardware.

Verdict

RAID 6 is the correct default for any array with four or more drives over 8 TB each. RAID 5 remains fine for small arrays (3-4 drives) with sub-4 TB disks, particularly on enterprise drives with 10^15 URE rates. Beyond those limits, the rebuild failure risk of RAID 5 makes RAID 6 the safer, mandatory for large arrays choice, even at the capacity cost.

Calculate your own RAID capacity

Which should you pick?

Choose RAID 5

Pick RAID 5 for 3-4 drive arrays under 20 TB total capacity, especially with enterprise drives rated at 10^15 URE or better. Acceptable for secondary / non-critical data when combined with a real backup.

Size a 3-4 drive array

Choose RAID 6

Pick RAID 6 for any array of 5+ drives, any array using drives 10 TB or larger, and any array holding irreplaceable data. The capacity loss is worth the double fault tolerance.

See 10 TB rebuild over 10 GbE

Related comparisons