RAID Calculator – How Much of Your Array You Can Actually Use
Four 4 TB drives promise 16 TB on the box — and then RAID takes its share. Parity, mirror copies and spare protection all eat capacity before a single file lands, and every level eats a different amount. Tell this calculator how many disks you have, how big each one is, and which RAID level you are weighing up: it shows the usable space in both TB and TiB, what the protection costs, and how many disk failures the array can survive — so the enclosure you buy matches the space you actually need.
Your disks and the level
🔒 Everything is calculated in your browser — your numbers never leave your device.
What the array gives you
The full picture
What this means
RAID 5 spreads your data across all 4 disks and spends the space of one disk — 4.00 TB here — on parity, the repair information that lets the array rebuild a dead disk from the survivors. You keep 12.00 TB to store on, and the array keeps running through any one disk failure. The price is written into the level itself: one disk of capacity, and slower writes while parity is worked out.
How to read the two units. Drive makers sell terabytes (TB): 1 TB is 1,000 GB. Operating systems count tebibytes (TiB) but often label them TB: 1 TiB is 1,024 GiB. The same bytes give two honest numbers — 12.00 TB and 10.91 TiB here — and the gap grows with size, which is why a big array always looks smaller on screen than on the invoice.
- Four 4 TB drives promise 16 TB on the box — and then RAID takes its share: parity, mirror copies and spare protection all eat capacity before a single file lands.
- On the default — four 4 TB disks — the usable space is 12.00 TB (10.91 TiB).
- The maths assumes every disk is the same size: mixing sizes is allowed by most controllers, but then every disk counts only as much as the smallest one.
Every level, on the same disks
The comparison that decides most purchases. Same disks, six ways to arrange them: watch usable space fall as protection rises, and notice that the two mirror-heavy levels cost half the array. The highlighted row is the level picked above — change the disk count or size and the whole table follows.
| RAID level | Raw | Usable (TB) | Usable (TiB) | Efficiency | Disks it can lose |
|---|---|---|---|---|---|
| RAID 0 | 16.00 TB | 16.00 | 14.55 | 100.0% | None |
| RAID 1 | 16.00 TB | 4.00 | 3.64 | 25.0% | 3 |
| RAID 5 | 16.00 TB | 12.00 | 10.91 | 75.0% | 1 |
| RAID 6 | 16.00 TB | 8.00 | 7.28 | 50.0% | 2 |
| RAID 10 | 16.00 TB | 8.00 | 7.28 | 50.0% | 1 per pair |
| RAID 0+1 | 16.00 TB | 8.00 | 7.28 | 50.0% | 1 per copy |
Efficiency is usable space divided by raw space. RAID 1 with four disks mirrors the same data onto every disk, so it reads 25% here — most people run RAID 1 as a pair, where it reads 50%. Levels greyed out in the live table need more disks, or an even count, than the array above them describes.
How it works, in 3 steps
Count the disks, honestly
Enter how many disks will actually sit in the array and the size printed on each one. Keep them identical if you can: mixed sizes drag every disk down to the smallest, so a bargain odd-sized spare can quietly shrink the whole array. The raw total is simply count times size — the number the shopping basket advertises.
Pick a level by what you fear
Each level is capacity traded for survival. RAID 0 trades everything for speed and keeps nothing back. Parity levels (5 and 6) keep a disk or two of repair information. Mirror levels (1, 10, 0+1) keep whole copies and charge half the array for them. Choose by the failure you need to live through, not by the biggest number.
Read both units before buying
Check the usable figure in TB against the drives you are buying, then check the TiB figure against what your operating system will display. If the TiB number disappoints, the fix is another disk or a bigger one — not a different calculator. Buy for the TiB you need and the TB argument settles itself.
How each figure is worked out
The default, worked through. Four 4 TB disks hold 16 TB raw. RAID 5 keeps one disk’s worth of parity, so the disks that hold data number 4 − 1 = 3, and usable capacity is 3 × 4 TB = 12 TB. RAID 6 keeps two disks’ worth: (4 − 2) × 4 TB = 8 TB. RAID 10 mirrors every disk once, so half the array holds data: (4 ÷ 2) × 4 TB = 8 TB. RAID 0 keeps no protection at all and uses everything: 16 TB. And a RAID 1 pair — two 4 TB disks mirroring each other — gives 4 TB. Convert with the factor above and 12 TB reads as 12 × 0.90949 = 10.91 TiB.
What parity actually is. Parity is repair arithmetic stored across the disks: for each stripe of data blocks, one block holds a value computed from the others, so any single missing block can be recomputed from the survivors. That is why RAID 5 loses exactly one disk of capacity whatever the array size — the parity blocks rotate across all the disks but always add up to one disk — and why RAID 6, carrying two independent parity values per stripe, loses exactly two disks and survives two failures.
Why mirrors charge half. A mirror is not arithmetic but a second copy, byte for byte. RAID 10 stripes data across pairs of mirrored disks, so every byte is stored twice and exactly half the raw space is usable — in exchange, a rebuild is a straight copy from the surviving twin, fast and gentle on the other disks, which is why busy servers so often accept the 50% charge.
Deep dives
Nothing is missing — the rulers differ. A terabyte, as drive makers and the SI system define it, is 1,000 × 1,000 × 1,000 × 1,000 bytes. A tebibyte is 1,024 × 1,024 × 1,024 × 1,024 bytes, about 9.95% more. Your operating system measures the second and, on many systems, mislabels it with the first unit’s name. So a 4 TB drive quite truthfully contains 4,000,000,000,000 bytes, and the computer quite truthfully reports 3.64 of its binary units. Both are right; they are counting different-sized buckets.
The gap compounds with size. At gigabyte scale the two rulers differ by about 7.4%, at terabyte scale by about 9.95%, and at petabyte scale by about 12.6% — the ratio multiplies at every step because 1,024 ÷ 1,000 is raised to a higher power each time. That is why the disappointment grows with the array: a single small drive looks a few percent short, while this page’s 12 TB array reads 1.09 TiB lower than its TB figure, and a 100 TB array would read about 9 TiB lower.
How to plan with both numbers. Buy disks using TB, because that is how they are sold, and budget your folders using TiB, because that is how they will be measured. If a backup target must hold a 10 TiB working set, the drives must sum — after RAID has taken its share — to at least 10 ÷ 0.90949, about 11 TB of usable decimal capacity. This calculator shows both figures side by side precisely so that conversion never happens in your head at the worst moment.
The rebuild is the dangerous moment. When a disk dies, the array rebuilds it by reading every surviving disk end to end. On modern multi-terabyte drives that can take a day or more, and the survivors are old disks being worked at their hardest. If a second disk fails — or one unreadable sector turns up — a RAID 5 array is gone. That risk, small with small drives, became the standard argument for RAID 6 as drives grew: the second parity means the array survives a failure that happens during the rebuild itself.
Writes pay a tax on parity levels. Changing one block on RAID 5 means reading the old data and old parity, then writing new data and new parity — four operations for one change; RAID 6 pays six. Random-write-heavy work, like databases and virtual machines, feels that tax keenly. RAID 10 pays almost none — a write is just two copies written at once — and rebuilds by copying one disk instead of recomputing from all of them. Its 50% capacity charge buys the calmest failure behaviour of the three.
And none of them is a backup. RAID protects against a disk dying. It does nothing against deletion, ransomware, a controller mangling the whole array, fire, theft or flood — every copy and every parity block sits in the same box, sharing the same fate, and deletions replicate instantly and faithfully. Keep a separate copy of anything that matters, ideally somewhere else. RAID buys uptime; backups buy recovery; serious setups carry both.
Every level on the same disks: what each one actually costs you
Take the default array — four disks of 4 TB each, 16 TB of raw capacity — and run every level through this page’s own calculator. The table is the whole RAID trade in one view: usable space, what is handed to protection, and how many disks can fail before data is lost. Nothing here is a recommendation for a particular use; it is the arithmetic every recommendation has to start from.
| Level (4 × 4 TB) | Usable capacity | Handed to protection | Efficiency | Disk failures survived |
|---|---|---|---|---|
| RAID 0 — striping | 16.00 TB (14.55 TiB) | 0.00 TB | 100.0% | 0 — one failure takes the array |
| RAID 1 — mirroring | 4.00 TB (3.64 TiB) | 12.00 TB | 25.0% | 3 — every disk holds a complete copy |
| RAID 5 — single parity | 12.00 TB (10.91 TiB) | 4.00 TB | 75.0% | 1 — any single disk |
| RAID 6 — double parity | 8.00 TB (7.28 TiB) | 8.00 TB | 50.0% | 2 — any two disks at once |
| RAID 10 — mirrored pairs, striped | 8.00 TB (7.28 TiB) | 8.00 TB | 50.0% | 1 in each mirror pair |
| RAID 0+1 — striped sets, mirrored | 8.00 TB (7.28 TiB) | 8.00 TB | 50.0% | 1 in each striped copy |
Three rows deserve a second look, because each hides a common misunderstanding. First, RAID 1 on four disks gives only 4.00 TB of usable space, not 8.00 TB: this calculator mirrors the same data onto every disk, so every disk holds a complete copy and the array survives three failures — but it pays for that survival with three quarters of its raw capacity. A four-disk RAID 1 is, in effect, a single 4 TB volume with three spare copies of itself. Second, RAID 10 and RAID 0+1 give identical capacities on these disks (8.00 TB each) and identical efficiency (50.0%), but they do not survive failures identically: RAID 10 survives one failure in each mirror pair, so it can survive two failures if they land in different pairs, while RAID 0+1 survives one failure in each striped copy, with its own pattern of which second failures are fatal. Identical capacity tables do not mean identical protection, and the “failures survived” column is where the two levels actually differ. Third, RAID 0’s 100% efficiency is the most expensive row in the table in a different currency: a single disk failure destroys the whole array, because every byte is striped across all four disks and nothing is held back. Efficiency bought with zero redundancy is not a bargain; it is a gamble priced in data rather than in terabytes.
The parity levels reward larger arrays, and the calculator shows why. RAID 5 on three disks of 4 TB gives 8.00 TB usable at 66.7% efficiency — one disk of the three goes to parity. On four disks it gives 12.00 TB at 75.0%; on six disks of 6 TB it gives 30.00 TB at 83.3%; on eight disks of 8 TB it gives 56.00 TB at 87.5%. The parity cost is always exactly one disk, so its share shrinks as the array grows: 1/3, then 1/4, then 1/6, then 1/8. RAID 6’s cost is always two disks, so it shrinks the same way but from a lower start. That scaling is the real argument for RAID 5 and 6 on larger arrays — and the real argument against judging a level by its efficiency on a small array, where the parity overhead looks worst. A level that wastes a third of a three-disk array wastes an eighth of an eight-disk one, with the same single-disk protection in both cases.
Terabytes, tebibytes, and the missing 9%
Every row of the table above shows two capacities — for example, RAID 5 on the default array gives 12.00 TB and 10.91 TiB — and the gap between them is not a rounding error or a manufacturer trick in the sense most people mean. Disk makers count in decimal units: 1 TB is 1,000,000,000,000 bytes. Operating systems count in binary units: 1 TiB is 1,099,511,627,776 bytes (1024⁴). A terabyte is therefore only 0.9095 tebibytes, so 12.00 TB of usable decimal capacity reports as 10.91 TiB in the operating system, and 16.00 TB of raw capacity reports as 14.55 TiB. The “missing” 9% was never there in binary terms; it is the difference between two rulers, not a loss of storage. This calculator reports both figures for exactly that reason: the TB figure is what you bought, and the TiB figure is what your computer will show you.
The same ruler difference explains why a new array always looks smaller than its box promised, before any RAID overhead is counted. Four disks labelled 4 TB hold 16.00 TB decimal, which is 14.55 TiB binary — already 1.45 TiB “missing” before RAID takes anything. Then RAID 5 takes one disk of parity, leaving 12.00 TB, which is 10.91 TiB. A buyer who expected 16 TiB and sees 10.91 TiB has lost 1.45 TiB to units and 3.64 TiB to parity, and only the second loss is a real trade: the first is arithmetic. Filesystems take a further slice on top, for their own structures, which this calculator does not model — so even the TiB figure here is an upper bound on what will be free for files, not a promise of free space. When planning an array, size it in the units your operating system will report, add the filesystem’s share, and only then compare levels; comparing decimal box figures against binary free-space figures is how arrays end up a tenth smaller than planned before a single file is written.
One last scope note, because it is the most important sentence on any RAID page. RAID is not a backup, as this page’s own FAQ states, and no level in the table changes that. RAID protects against disk failure; it does not protect against deletion, overwriting, ransomware, theft, fire or a controller fault that corrupts the array itself — in every one of those cases, every copy RAID holds is damaged together, because RAID’s copies are not independent backups but parts of one live volume. The usable capacities above are capacities for live data with disk-failure protection built in. A backup is a separate copy, on separate media, ideally in a separate place, and it is needed at every level from RAID 0 to RAID 6 alike. Size the array with this calculator; protect the data with something else as well.
Rebuilds: the window when protection is thinnest
Usable capacity is only half the protection story; the other half is what happens after a disk actually fails. On RAID 5, the loss of one disk leaves the array running in a degraded state: every read that needs the missing disk’s data must be reconstructed from parity and the surviving disks, and there is no protection left until a replacement disk is fitted and the rebuild completes. A second failure during that window — or a read error on a surviving disk while it is being read hard for the rebuild — is the event RAID 5 cannot survive. That is the real cost of single parity on large arrays: the rebuild window grows with disk size, because more data must be read and rewritten, and the window is precisely when the array is most exposed. RAID 6 exists for this window: with two parities, the array survives a second failure during the rebuild of the first, which is why double parity is the usual choice where disks are large and rebuilds are long.
RAID 10 rebuilds differently, and the difference matters. When a disk fails in a mirrored pair, the rebuild copies the surviving mirror onto the replacement — a straight copy from one disk, rather than a reconstruction that reads every surviving disk. The array is exposed only in that one pair during the copy, and the other pairs keep their full protection. That is why RAID 10 is often described as rebuilding faster and more safely than parity RAID, even though its usable capacity (50.0% on the default array) is lower than RAID 5’s (75.0%). Capacity, rebuild behaviour and failure tolerance are three separate dials, and this calculator prices only the first. When choosing between levels, price all three: what the array holds, what it survives, and how it behaves in the hours after something has already gone wrong.
Four 4 TB disks, five answers: what each level actually gives you
Same hardware, wildly different usable space — the level is the decision, not the disks.
Four 4 TB disks hold 16 TB of raw storage, and no level changes that. What changes is how much of it you can use: RAID 0 stripes across all four for the full 16 TB with no protection; RAID 1 mirrors everything for 4 TB usable; RAID 5 holds back one disk for parity and gives 12 TB; RAID 6 holds back two and gives 8 TB; RAID 10 mirrors two striped pairs and also gives 8 TB. The 12 TB of a RAID 5 array is 10.91 TiB in the units operating systems report — 12 TB × 0.9095 — which is why a new array always looks about 9% smaller than the box promised.
Capacity is only half the trade. RAID 5’s parity cost is always exactly one disk, so its efficiency improves as arrays grow — one disk in four is 25% overhead, one in eight is 12.5% — while its rebuild window, the period when a second failure would be fatal, grows with disk size. RAID 6 pays a second disk for the right to survive two failures, which is why it is the default advice for large arrays of large disks. RAID is not backup in any of these forms: it protects against disk failure, not against deletion, ransomware, fire or theft, and every level here needs a separate backup to be a safe system.
Every level on the same disks, rebuild windows and the TiB gap: RAID levels explained.
| Level | Usable (4 × 4 TB) | Disks lost | Survives |
|---|---|---|---|
| RAID 0 | 16 TB | 0 | No disk failure |
| RAID 1 | 4 TB | 3 | All but one mirror |
| RAID 5 | 12 TB | 1 | One disk |
| RAID 6 | 8 TB | 2 | Two disks |
| RAID 10 | 8 TB | 2 | One per mirror pair |
The rebuild window: when protection is thinnest
Rebuilding stresses every surviving disk. The risk is not the failed disk — it is the next one.
When a disk fails, the array rebuilds onto a replacement by reading every surviving disk end to end. On multi-terabyte disks that takes hours or days, and every surviving disk is worked at full tilt for the whole of it — the worst possible moment to discover that a second disk has a latent defect. That is the rebuild window, and it is the argument for RAID 6 on big arrays: two parity disks mean the array can absorb a second failure mid-rebuild and keep going, where RAID 5 cannot.
Three habits shrink the window’s danger. Keep a hot or cold spare so the rebuild starts at once instead of waiting for a delivery. Watch the array’s health reports — a disk throwing errors before it fails is a warning, not a nuisance. And never treat the rebuild as routine: backups should be verified before the rebuild starts, because the rebuild is precisely when you are most likely to need them. The calculator’s capacity figures assume healthy disks; the rebuild is where that assumption gets tested.
RAID questions, answered
What is a RAID calculator?
A RAID calculator works out how much usable storage a set of disks gives you under a chosen RAID level, before you build the array. Feed it the number of disks, the size of each one and the level — 0, 1, 5, 6, 10 or 0+1 — and it applies that level’s capacity rule: all the disks for RAID 0, one disk held back for parity on RAID 5, two on RAID 6, half the array for mirrors on RAID 10 and 0+1. It also shows the raw total, the space handed to protection, the storage efficiency, and how many disk failures the array can survive, in both TB and TiB so the figure matches what an operating system will report.
How much usable space does RAID 5 give?
RAID 5 gives you the capacity of all your disks minus one: usable space is (number of disks − 1) × disk size. Four 4 TB disks give (4 − 1) × 4 TB = 12 TB usable — 10.91 TiB as an operating system counts it — while the remaining 4 TB holds parity, the repair information spread across the array. The rule scales kindly: eight 4 TB disks give 28 TB usable, so the parity cost falls from a third of a small array to an eighth of a larger one. RAID 5 needs at least three disks, and the array keeps running through any one disk failure.
Why does my array show less space than the drives promise?
Two honest effects stack up. First, RAID itself: parity or mirror copies take their share before you store anything, which this calculator shows as the handed-to-protection figure. Second, units: drives are sold in terabytes (1 TB = 1,000 GB) while operating systems measure tebibytes (1 TiB = 1,024 GiB) and often label them TB anyway. A 12 TB usable array therefore appears as 10.91 on screen. Nothing was taken by the manufacturer or the filesystem — the same bytes are being counted with two different rulers, and the difference is about 9.95% at terabyte scale.
What is the difference between RAID 5 and RAID 6?
RAID 6 is RAID 5 with a second, independent parity: it keeps two disks’ worth of repair information instead of one, so it survives any two disks failing at once — including a second failure during the long rebuild after the first. The price is capacity and write speed: four 4 TB disks give 8 TB usable under RAID 6 against 12 TB under RAID 5, and each write costs more parity work. RAID 6 needs at least four disks. With today’s multi-terabyte drives and day-long rebuilds, RAID 6 is the usual recommendation for arrays that hold data you cannot re-create; RAID 5 remains reasonable for small disks and data that is backed up elsewhere.
How many disks can fail in RAID 10?
RAID 10 stripes data across mirrored pairs, so the honest answer is: at least one disk, and up to one disk per pair — provided the failures land in different pairs. An eight-disk RAID 10 array is four pairs; it survives one failure in each pair, four unlucky disks in total, but losing both members of any single pair loses the array. That is the mirror bargain: exactly half the raw capacity is usable — eight 4 TB disks give 16 TB — in exchange for fast rebuilds, which copy one surviving twin instead of recomputing from every disk, and strong write performance. RAID 10 needs an even number of disks, at least four.
Is RAID a backup?
No. RAID protects against one specific accident — a disk failing — and keeps the array running through it. It offers no protection against deleting the wrong folder, ransomware encrypting the volume, a controller corrupting every disk at once, or the box being stolen, flooded or burned, because the copies and parity all live in the same machine and deletions are mirrored instantly and faithfully. A backup is a separate copy, kept apart from the array, ideally in another location and not permanently connected. Use RAID for uptime, backups for recovery, and both for anything you genuinely cannot afford to lose.
Why does RAID 1 on four disks give only 4 TB?
Because this calculator mirrors the same data onto every disk: every disk holds a complete copy, so the usable space is the size of one disk — 4.00 TB — however many disks there are, and the array survives three failures. Four-disk RAID 1 is a single 4 TB volume with three spare copies of itself. If you want 8 TB usable from four 4 TB disks with mirroring, you want RAID 10 or RAID 0+1, which mirror pairs or sets rather than every disk.
RAID 10 and RAID 0+1 give the same capacity. Are they the same?
No. On four 4 TB disks both give 8.00 TB usable at 50.0% efficiency, but they survive failures differently: RAID 10 survives one failure in each mirror pair (so two failures if they land in different pairs), while RAID 0+1 survives one failure in each striped copy, with a different pattern of which second failures are fatal. Identical capacity tables do not mean identical protection — read the “failures survived” line, not just the usable figure.
Why does my operating system show 10.91 TiB for a 12 TB array?
Because the two figures use different units. Disk makers count 1 TB as 1,000,000,000,000 bytes; operating systems count 1 TiB as 1,099,511,627,776 bytes. One terabyte is only 0.9095 tebibytes, so 12.00 TB is 10.91 TiB. Nothing is lost: the same bytes measured with a larger ruler give a smaller number. This calculator shows both figures so the box capacity and the on-screen capacity can be compared without confusion.
Does RAID 5 get more efficient on bigger arrays?
Yes, because the parity cost is always exactly one disk. On three 4 TB disks RAID 5 gives 8.00 TB at 66.7%; on four it gives 12.00 TB at 75.0%; on six 6 TB disks it gives 30.00 TB at 83.3%; on eight 8 TB disks it gives 56.00 TB at 87.5%. The protection — surviving any single disk — is the same in every case; only the overhead share shrinks as the array grows.
Why does my 12 TB RAID 5 array show as 10.91 TiB?
Because disk makers and operating systems count differently. Makers count a terabyte as 1,000,000,000,000 bytes; operating systems count a tebibyte as 1,099,511,627,776 bytes. The same 12 trillion bytes are 12 TB in the maker’s units and 10.91 TiB in the system’s — about 9% “missing” that was never there. The calculator reports both, so the quote and the screen can be compared without confusion.
Does RAID replace backups?
No. RAID protects against disk failure only. It does not protect against accidental deletion, ransomware, fire, theft or a failed controller — all of which take the whole array, redundancy included. Every RAID level needs a separate backup, stored elsewhere, to be a safe system. Treat RAID as uptime insurance and backups as data insurance; neither substitutes for the other.
Related calculators
IP Subnet Calculator
Network ranges, hosts and masks — the other address-space puzzle.
🔁Conversion Calculator
Everyday unit conversions in one place, GB to TB included.
％Percentage Calculator
Shares, increases and discounts — the quick way, no algebra.
🧮Scientific Calculator
Powers, roots and functions — check any figure by hand.
🎲Random Number Generator
Pick numbers in any range, with or without repeats.
💾RAID Levels Explained
Capacity, parity, rebuilds and the TB/TiB gap for every level.
Standard RAID maths on your figures — read the fine print of your controller.
The figures assume equal-sized disks and the classic form of each level. Real controllers and NAS systems add variations: hot spares held out of the array, hybrid systems like SHR that mix disk sizes more cleverly, filesystem overhead, and vendor-specific rebuild behaviour all move the final number a little. Treat this page as the planning figure — the capacity the level itself allows — and confirm the exact free space on your own hardware once the array is built. And whatever the level: RAID is not a backup, so keep a separate copy of what matters.