RAID Calculator
Usable capacity for RAID 0, 1, 5, 6, 10, 50, 60 and JBOD, worked from each level's definition rather than looked up, with the same disks costed under every level side by side. Fault tolerance is shown as the number that survives whichever disks fail, with the luckier number beside it rather than instead of it, and a TB/TiB switch converts the answer into the figure Windows, Linux and most NAS software will report.
RAID 5 · 4 disks · 4 TB eachusable of 16 TB raw · any 1 disk can fail, whichever one it is.
Usable 12 TB (75%)Parity 4 TB (25%)
Usable = (4 − 1) × 4 TB = 12 TB
A tebibyte is exactly 1,099,511,627,776 bytes, so 12 TB of usable capacity is 10.91 TiB. Windows, Linux and most NAS software count that way and then print the figure as TB anyway, which is why a finished array looks smaller than the invoice.
Mixed disk sizes: every level except JBOD treats all disks as the size of the smallest one, so a 6 TB disk in an array of 4 TB disks contributes 4 TB. Enter the smallest disk's size.
The same disks under every level
4 disks of 4 TB, costed under each level. Pick a level name to switch to it. RAID 50 and 60 use the grouping that keeps the most capacity.
| Level | Usable | Lost to protection | Disks that can fail | Minimum disks |
|---|---|---|---|---|
| 16 TB | 0 TB (0%) | 0 | 2 | |
| 4 TB | 12 TB (75%) | 3 | 2 | |
| 12 TB | 4 TB (25%) | 1 | 3 | |
| 8 TB | 8 TB (50%) | 2 | 4 | |
| 8 TB | 8 TB (50%) | 1 (up to 2) | 4 | |
| RAID 50 needs at least 6 disks and you have 4. It stripes across RAID 5 groups, and the smallest group is three disks. Add 2 more disks, or choose RAID 5. | 6 | |||
| RAID 60 needs at least 8 disks and you have 4. It stripes across RAID 6 groups, and the smallest group is four disks. Add 4 more disks, or choose RAID 6. | 8 | |||
| 16 TB | 0 TB (0%) | 0 | 1 | |
RAID is not a backup. It keeps an array serving through a disk failure, and that is all it does. A deleted file, a ransomware run, a bad controller, a fire or a theft reaches every disk at once, because every disk is in the same box on the same power supply. Keep a copy somewhere else as well.
RAID levels compared
N is the number of disks. Usable capacity assumes disks of one size, and every level except JBOD uses the smallest disk in the array as that size.
| Level | What it does | Minimum disks | Usable capacity | Disks that can fail | Reads and writes |
|---|---|---|---|---|---|
| RAID 0 | Striping, no redundancy | 2 | N × the smallest disk | None | Reads and writes both stripe across every disk: the fastest level, and the only one where a single failure loses everything. |
| RAID 1 | Mirroring | 2 | One disk | All but one disk | Reads can be served by any copy. Writes go to every disk, so writing is as fast as one disk and no faster. |
| RAID 5 | Striping with distributed parity | 3 | (N − 1) × the smallest disk | 1 disk | Fast reads. A small write costs a read and a write of the parity as well, which is the read-modify-write penalty. |
| RAID 6 | Striping with double parity | 4 | (N − 2) × the smallest disk | 2 disks | Fast reads. Every write updates two parity blocks, so small writes cost more than RAID 5. |
| RAID 10 | Mirrored pairs, striped | 4 | N ÷ 2 × the smallest disk | 1 disk, and up to one in every pair | Fast reads and fast writes, with no parity to compute. The usual choice under a busy database. |
| RAID 50 | Striped RAID 5 groups | 6 | (N − groups) × the smallest disk | 1 disk in every group | Reads and writes behave like RAID 5, spread over the groups. Only the failed group rebuilds, so a rebuild is shorter and safer than one wide RAID 5. |
| RAID 60 | Striped RAID 6 groups | 8 | (N − 2 × groups) × the smallest disk | 2 disks in every group | Reads and writes behave like RAID 6, spread over the groups. The safest striped parity there is, and the most expensive in capacity. |
| JBOD | Just a bunch of disks | 1 | Every disk in full | None, but a failure costs only that disk | No striping, so a read or a write runs at the speed of whichever single disk holds the file. |
Common questions
- What is the difference between RAID 5 and RAID 6?
- One parity disk against two. RAID 5 gives up one disk's worth of capacity across the whole array and survives any single failure. RAID 6 gives up two and survives any two. On four 4 TB disks that is 12 TB usable against 8 TB. The extra disk buys safety during a rebuild: replacing a failed disk in a RAID 5 array means reading every remaining disk end to end, and a second failure or one unreadable sector during that read loses the array, while RAID 6 still has a parity block in hand. The bigger the disks, the longer that window stays open, which is why double parity is the common recommendation on large drives.
- How much usable space does RAID 10 give?
- Exactly half. RAID 10 mirrors disks in pairs and stripes across the pairs, so eight 4 TB disks give 16 TB usable out of 32 TB raw. It needs an even number of disks and at least four. What the other half buys is speed, since there is no parity to compute, and a short rebuild, since recovering a failed disk copies its mirror instead of reading every other disk. Fault tolerance is one disk guaranteed and up to one per pair if the failures fall kindly, so this calculator prints both numbers rather than the flattering one.
- Why does a 12 TB array show up as 10.9 TB?
- Because two different units share one abbreviation. Disks are sold in decimal terabytes, where 1 TB is 1,000,000,000,000 bytes. Windows, Linux and most NAS software count in binary, where the unit is a tebibyte of exactly 1,099,511,627,776 bytes, and then usually label it TB anyway. The bytes are identical and only the divisor changes: 12 divided by 1.099511627776 is 10.91. The TB and TiB switch here shows the answer either way, so the number you compare against the operating system and the number on the invoice are on the same page.
- What happens if the disks are different sizes?
- Every level except JBOD treats all disks as the size of the smallest one, because a stripe or a mirror has to be the same width on each disk. A 6 TB disk in an array of 4 TB disks contributes 4 TB and the other 2 TB is stranded. Enter the smallest disk's size for a true answer. JBOD is the exception: it does not stripe, so each disk is used in full and the total is the sum of the real sizes, at the cost of no redundancy at all.
- What are RAID 50 and RAID 60?
- Striped sets of smaller parity groups. RAID 50 splits the disks into equal RAID 5 groups and stripes across them, so twelve disks in three groups of four give up three disks to parity rather than one, and survive one failure in each group rather than one in the whole array. RAID 60 does the same with RAID 6 groups and gives up two per group. The group count changes both the capacity and the tolerance, so it is a control here rather than an assumption: with twelve disks a RAID 50 can run as two, three or four groups, and each choice is priced here.
- Is RAID a backup?
- No. RAID keeps an array serving through a disk failure, and that is the whole of what it does. A deleted file is deleted on every disk at once, and so is a ransomware run, a bad controller write, a fire or a theft, because every disk sits in one box on one power supply. A backup is a second copy somewhere else, ideally somewhere the array itself cannot write to. This calculator prices the redundancy, not the backup.
Usable capacity follows each level's definition exactly: RAID 5 loses one disk, RAID 6 two, RAID 10 half. Real arrays lose a little more to metadata and filesystem overhead, and a rebuild's risk depends on disk size and error rate, which this does not model.