Work out usable storage capacity from your disks, or find out how many drives you need for a target capacity — for every common RAID level.
Each disk's inner bar fills a little more with every write it receives, and stays full (●) once at capacity.
RAID (Redundant Array of Independent Disks) combines multiple physical drives into one logical volume to improve performance, add redundancy, or both. The usable capacity you end up with is almost never the sum of your raw disk sizes — parity, mirroring and striping all take a cut in exchange for speed or fault tolerance. Getting the numbers right before you buy hardware avoids under-provisioning a server or overspending on drives you don't need.
Striping only. Full raw capacity and maximum speed, but zero fault tolerance — one failed disk loses everything.
Full mirroring. Usable capacity equals a single disk no matter how many drives are mirrored together.
Single distributed parity. Survives one disk failure. Needs at least 3 disks.
Double distributed parity. Survives two simultaneous disk failures. Needs at least 4 disks.
Mirrored pairs striped together. Strong performance and per-pair fault tolerance. Needs at least 4 disks, even count.
Multiple RAID 5 groups striped together. Balances capacity and redundancy on larger arrays.
Multiple RAID 6 groups striped together. Higher redundancy for large arrays at a higher disk cost.
Disks simply concatenated. No redundancy, no speed gain, just combined raw capacity.
Usable capacity depends on the RAID level: RAID 0 and JBOD give the full raw capacity, RAID 1 gives you one disk's worth regardless of drive count, RAID 5 gives (n-1) disks worth, RAID 6 gives (n-2), and RAID 10 gives half the raw capacity. This calculator applies the right formula automatically once you pick a level.
Switch to "I need capacity", enter your target usable capacity and disk size, and the calculator works out the minimum number of drives for the selected RAID level, respecting the minimum disk count each level requires.
RAID 5 tolerates one failed disk with a minimum of three disks. RAID 6 tolerates two simultaneous failures with a minimum of four disks, at the cost of one extra disk of usable capacity compared with RAID 5 on the same disk count.
RAID 5 and RAID 6 are common choices for balancing usable capacity with redundancy on medium arrays. RAID 10 is preferred when write performance and rebuild speed matter more than raw capacity efficiency. RAID 50 and RAID 60 extend that balance to larger arrays.
Rebuild time is roughly the size of the replacement disk divided by the throughput the array can sustain while it is still serving traffic. Pick your disk type and expected load and the calculator estimates it: a 16 TB 7200 rpm disk in a busy RAID 5 can take several days, while the same rebuild on NVMe finishes in under an hour. The RAID level matters less for the duration than for the risk. RAID 5 must read every surviving disk end to end, so a single unrecoverable read error during that window ends the rebuild, whereas RAID 6 still has a second parity to cover it.