RAID calculator
Choose a RAID level, the number of disks and their size to see how much space you can actually use, how many disks can fail without losing data, and roughly how fast reads and writes are compared with one disk.
Runs in your browser: nothing you enter is sent anywhere.
How to use it
- Choose the RAID level.
- Enter the number of disks and the size of one disk in TB (0.5 for a 500 GB disk). For RAID 50 and 60, also enter the number of groups.
- Read the usable capacity in TB and in TiB, the share of raw space you keep, the failures the array survives and the speed factors.
How it works
| Level |
Minimum disks |
Usable |
Survives always |
| RAID 0 |
2 |
n × s |
0 |
| RAID 1 |
2 |
s |
n − 1 |
| RAID 5 |
3 |
(n − 1) × s |
1 |
| RAID 6 |
4 |
(n − 2) × s |
2 |
| RAID 10 |
4, even |
n/2 × s |
1 (up to n/2) |
| RAID 50 |
6, 3 per group |
(n − g) × s |
1 (up to g) |
| RAID 60 |
8, 4 per group |
(n − 2g) × s |
2 (up to 2g) |
Speed factors use the write-penalty model: read n×, write n ÷ penalty ×. They compare small random operations with one disk of the same kind. Real numbers depend on the controller, cache, disk type and workload.
Worked example
RAID 5 with 4 disks of 4 TB:
- raw capacity 4 × 4 = 16 TB;
- usable (4 − 1) × 4 = 12 TB, which is 12 × 10^12 ÷ 2^40 = 10.91 TiB;
- efficiency 12 ÷ 16 = 75 %;
- it survives 1 failed disk;
- reads about 4× one disk, writes 4 ÷ 4 = 1× one disk.
The same 4 disks in RAID 10 give 8 TB (50 %) and write twice as fast as one disk (4 ÷ 2 = 2×), while still surviving 1 failure, or 2 if they are in different pairs.
Tips
- With large disks, a RAID 5 rebuild takes many hours, and a second failure during it loses the array; RAID 6 or RAID 10 is safer for big arrays.
- Plan for a hot spare: it is not counted here, so subtract it from the number of disks.
- Convert the result between TB, TiB and other units with the data storage converter.
Frequently asked questions
How is usable capacity calculated?
With n disks of size s: RAID 0 gives n × s, RAID 1 gives s (every disk holds the same data), RAID 5 (n − 1) × s, RAID 6 (n − 2) × s, RAID 10 n/2 × s, RAID 50 (n − g) × s and RAID 60 (n − 2g) × s, where g is the number of groups. Mixed disk sizes are not modelled: arrays use the smallest disk's size for every member.
Why does the OS show less space than the calculator?
Disk makers count 1 TB as 10^12 bytes, while Windows and many tools count in TiB, 2^40 bytes, and still write TB. 12 TB is 10.91 TiB. File system overhead takes a little more.
What do the two fault tolerance numbers mean?
Always survived is the number of failed disks the array is guaranteed to survive, whichever disks fail. At best is the most it can survive if failures land in the right places: in RAID 10 one disk per mirrored pair, in RAID 50/60 one or two per group.
How are the speed factors estimated?
With the common model for random input/output: reads can use every disk, so n times one disk; each write costs extra disk operations (the write penalty): 2 for RAID 10, n for an n-way RAID 1, 4 for RAID 5 and 50, 6 for RAID 6 and 60. Write speed is n divided by that penalty. Controller cache, sequential transfers, SSD behaviour and rebuilds are not modelled.
Is RAID a backup?
No. RAID keeps the array running when a disk fails, but deleted files, ransomware, controller faults and fires affect every disk at once. Keep separate backups.