RAID 6
RAID 6 Quick Facts
What Is RAID 6?
RAID 6 is a striped RAID layout that stores two independent sets of distributed parity information across the member drives. Compared with RAID 5, the additional parity allows the array to remain recoverable after the failure of up to two member drives.
The increased redundancy reduces nominal usable capacity by the equivalent of two member drives and adds more parity-maintenance work during writes, but it also provides additional protection during degraded operation and rebuilds.
RAID 6 Distributes Two Sets of Parity
Data and parity are distributed across the participating drives. Two independent parity calculations provide enough redundant information for the array to recover from the loss of any two members, assuming no additional failures or unrecoverable errors prevent reconstruction.
Simplified conceptual illustration of distributed dual parity.
RAID 6 Usable Capacity
RAID 6 consumes the equivalent capacity of two member drives for dual parity. As with other traditional RAID layouts, mixed-size arrays are generally limited by the smallest participating member's usable capacity.
Four 4 TB Drives
(4 − 2) × 4 TB = 8 TB nominal RAID 6 capacity .
Eight 8 TB Drives
(8 − 2) × 8 TB = 48 TB nominal RAID 6 capacity .
RAID 6 Can Tolerate Two Drive Failures
RAID 6 can reconstruct the array after the loss of up to two member drives. This provides an additional layer of protection compared with RAID 5, especially while an array is degraded.
A Third Member Failure Can Cause Array Failure
RAID 6 has finite redundancy. If a third required member becomes unavailable before sufficient redundancy is restored, the array may no longer contain enough information to recover all missing data.
The Main Difference Is Single Parity vs Dual Parity
RAID 5 uses one parity set and can tolerate one member-drive failure. RAID 6 uses two parity sets and can tolerate two. That additional protection costs the equivalent capacity of one more member drive and increases parity work during writes.
Read Performance vs Write Performance
Reads can benefit from striping across multiple active drives, particularly in sequential workloads. Real performance depends on the media, controller, interface, queue depth, workload, cache, stripe size, and software implementation.
Writes require maintaining two independent parity values. Small random writes can therefore involve more work than RAID 5, RAID 1, or RAID 10. Hardware RAID cache and workload characteristics can significantly change observed performance.
RAID 6 Rebuilds
When a failed member is replaced, the array reconstructs the missing content using surviving data and parity. Full protection is restored only after the required rebuild process completes successfully.
RAID 6 Retains Protection After the First Failure
With one failed member, RAID 6 still retains one additional drive-failure tolerance. This can be valuable in large arrays or during lengthy rebuild windows where a second failure is a concern.
RAID 6 Strengths
- Can tolerate up to two member-drive failures.
- Retains one-drive fault tolerance after the first failure.
- More capacity-efficient than mirroring in many larger arrays.
- Can provide strong aggregate read performance.
- Well suited to workloads where rebuild protection is important.
RAID 6 Limitations
- Consumes the equivalent capacity of two drives for parity.
- More parity write overhead than RAID 5.
- Rebuilds can remain lengthy on large-capacity drives.
- Latency-sensitive random writes may favor other layouts.
- RAID 6 does not replace independent backups.
When RAID 6 May Be Appropriate
RAID 6 is often considered when higher fault tolerance is more important than maximizing write performance or raw capacity.
RAID 6 Compared With RAID 5 & RAID 10
| RAID Level | Technique | Nominal Capacity | Fault Tolerance | Primary Tradeoff |
|---|---|---|---|---|
| RAID 5 | Single parity | (N − 1) × smallest drive | 1 drive | Higher efficiency, less fault tolerance |
| RAID 6 | Dual parity | (N − 2) × smallest drive | 2 drives | Higher redundancy with more parity overhead |
| RAID 10 | Mirrors + striping | Typically 50% | Topology-dependent | Strong write performance with lower capacity efficiency |
Calculate Your RAID 6 Configuration
Use the Tech Supply Direct RAID Calculator to estimate RAID 6 capacity, storage efficiency, fault tolerance, hot-spare impact, estimated performance, and hardware cost.
RAID 6 FAQ
How many drives are required for RAID 6?
RAID 6 requires a minimum of four participating drives.
How many drives can RAID 6 lose?
RAID 6 can tolerate up to two member-drive failures before its parity protection is exhausted.
How much capacity does RAID 6 provide?
Nominal usable capacity is generally calculated as the number of drives minus two, multiplied by the capacity of the smallest participating member.
Is RAID 6 safer than RAID 5?
RAID 6 provides greater drive-failure tolerance because it can survive two member failures instead of one, although it uses more capacity and parity-processing resources.
Is RAID 6 slower than RAID 5?
Some write workloads can incur additional overhead because RAID 6 maintains two parity values. Actual performance depends heavily on the controller, media, cache, workload, and implementation.
Is RAID 6 a backup?
No. RAID 6 protects against supported drive failures but does not replace an independent, tested backup strategy.
RAID 6 Redundancy Is Not a Backup
Dual parity protects against supported member-drive failures, but it does not protect against every cause of data loss. Accidental deletion, corruption, ransomware, controller problems, catastrophic hardware loss, and other events can still affect the array. Important data should be protected with independent backups.