RAID Performance
Estimate read/write performance for different RAID levels and drive counts.
About RAID Performance
RAID read performance scales linearly with disk count since reads can be striped across all drives, but write performance is reduced by a RAID-level-specific write penalty. Parity RAID (5, 6) pays a read-modify-write penalty to recalculate and rewrite parity blocks, while mirrored RAID (10) pays a smaller, fixed 2x penalty for writing to both members of each mirror pair. A workload's blended IOPS depends on its read/write mix, since writes consume disproportionately more back-end disk operations.
Formulas
Read IOPS
read_IOPS = disks × single_disk_IOPS
Write IOPS
write_IOPS = (disks × single_disk_IOPS) / write_penalty
- P_{write}
- 4 for RAID 5, 6 for RAID 6, 2 for RAID 10
Worked example
With the default values below, RAID Performance returns Blended IOPS (given read/write mix) of 632. Change any input above and every figure here is recalculated in your browser.
Inputs used
- Number of Disksminimum 2
- 8
- Single Disk IOPSminimum 1
- 150
- RAID Levelone of RAID 5, RAID 6, RAID 10
- RAID 5
- Read % of Workloadaccepts 0–100
- 70
Results
- Blended IOPS (given read/write mix)
- 632
- Max Read IOPS
- 1,200
- Max Write IOPS
- 300
- RAID Write Penalty
- 4
Step by step
- 1Values used
- 2Read IOPS
- 3Write IOPS
- 4Blended IOPS (given read/write mix)= 632
- 5Max Read IOPS= 1,200
- 6Max Write IOPS= 300
- 7RAID Write Penalty= 4
Assumptions and limits
- The result is derived only from the 4 inputs listed above. Anything not among them is not part of the calculation.
- In particular, this calculator does not take inflation, tax or compounding frequency into account.
- Figures are rounded for display to whole numbers, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.
How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.
Frequently Asked Questions
Why is the RAID 5 write penalty 4 and RAID 6 is 6?
RAID 5 requires 4 back-end I/Os per logical write: read old data, read old parity, write new data, write new parity. RAID 6 requires 6, since it maintains two independent parity blocks: read old data, read both old parities, write new data, write both new parities.
Why is RAID 10's write penalty only 2 despite also being redundant?
RAID 10 writes are simple duplication — each logical write becomes two physical writes (to the primary and its mirror) with no parity calculation or read-before-write step, making it far more write-efficient than parity RAID at the cost of using half the raw capacity for redundancy.
Does adding more disks always increase blended IOPS?
Yes for a fixed RAID level and workload mix, since both read and write IOPS scale with disk count — but the relative benefit is workload-dependent: read-heavy workloads gain IOPS almost linearly, while write-heavy workloads on parity RAID gain more slowly because each added disk's IOPS is still divided by the same write penalty.