What Is the RAID Calculator?
This RAID calculator will eliminate any confusion you have about which RAID level to choose (RAID — Redundant Array of Inexpensive/Independent Disks). It explains each of the commonly used RAID configurations and compares them on storage size, RAID performance, fault tolerance and cost, so you can make an informed choice about the configuration of your next RAID array.
It compares the following RAID levels side by side, using the same drives you entered: RAID 0, RAID 1, RAID 1E, RAID 10, RAID 5, RAID 50, RAID 5E, RAID 5EE, RAID 6 and RAID 60.
Capacities are shown in both unit systems at once: decimal (SI) — GB and TB, powers of 1000, the way drives are advertised and sold — and binary (IEC) — GiB and TiB, powers of 1024, the way Windows, Linux and most RAID controllers actually report free space. This is why a "4 TB" drive shows up as 3.64 TiB once it is formatted: nothing was lost, the two systems just count differently.
What Is a RAID Array?
In the early days of computing, mainframes used large and expensive hard disks designed to be highly reliable. The thing is that when they fail — and everything fails eventually — all of the data is lost unless you have a backup, and the expensive disk has to be replaced. Soon cheaper hard drives came along, but they were not very reliable at all, with failures being all too common. What was the solution? Use many disks together, acting as one.
RAID allows the creation of a single logical disk — it looks like one disk from the computer's point of view — made up of many cheap hard drives. There are many ways to arrange and configure those disks, depending on whether you want high performance (e.g., for video editing), high reliability, low cost, or something in between. These arrangements are known as RAID levels.
Please note that whatever RAID configuration you choose, you still need to back up the data. Preferably to magnetic tape, but online remote backup or a second RAID array will do. Just make sure there is a complete copy of the data somewhere else. Ideally, two copies. RAID protects against a drive failing; it does nothing about fire, theft, ransomware or a deleted folder.
RAID 0 — Striping (RAID 0 Calculator)
RAID 0 is all about performance. Data is divided into blocks and written alternately to each drive: odd-numbered blocks to disk 0, even blocks to disk 1. This process is called striping. Both writing and reading happen simultaneously across the drives, so the performance is double that of a single disk with two drives — and with three drives in RAID 0 you get 3× the read and write performance. You are only limited by the maximum number of disks you can connect to a single RAID controller.
There is a major downside to this stellar performance. If even one disk fails in a RAID 0 array, you lose all your data. RAID 0 is excellent where you need maximum performance (e.g., gaming, scratch disks for video editing) and the data exists somewhere else, such as a more fault-tolerant array or on tape.
- Usable capacity: n × C — nothing is held back.
- Fault tolerance: 0 drives.
- Minimum drives: 2.
RAID 1 — Mirroring (RAID 1 Calculator)
In contrast to RAID 0, RAID 1 is all about fault tolerance and reliability. The data is mirrored onto each drive, so even if one drive fails you still have all your data on the other. The write performance is the same as writing to one disk, since the controller writes the same data to both drives simultaneously. For reading, in theory the maximum speed is the sum of the speeds of the drives; in practice it is usually considerably slower, so choose your RAID hardware and software carefully to maximize read performance.
You might think to add more disks to a RAID 1 array, but this dramatically increases the cost per usable terabyte. With three mirrored drives, two disks could fail without data loss — but the usable size is only a third of the total capacity. Other RAID levels offer excellent reliability at less cost.
- Usable capacity: C — one drive, no matter how many you add.
- Fault tolerance: n − 1 drives.
- Minimum drives: 2.
What Is RAID 5? — Parity (RAID 5 Calculator)
A RAID 5 array is similar to RAID 0 in that data is striped across several disks. However, to make it fault tolerant, an extra block called a parity block is written for every stripe, and those parity blocks are spread across all the drives. Therefore the minimum number of drives is 2 + 1 — two for the actual data and one drive's worth for the parity.
So how does parity help? Say the data labeled A is split into A1, A2, A3 plus a parity block Ap. If disk 1 fails, A2 is lost — but it can be recovered in a process called rebuilding, because the parity block is calculated so that all the data blocks plus the parity block XOR together to zero. That means A2 can be computed from A1, A3 and Ap.
Rebuilding a RAID array takes time and a significant amount of processing power. It reads all the data from the remaining disks, calculates the lost data, and writes the recovered data back to a replacement disk. It does not scale well with today's enormous drives and can take hours, if not days. During the rebuild the array is vulnerable to a second disk failure, which would mean total loss of data — this is the single biggest argument against RAID 5 on large modern drives.
In terms of performance, a RAID 5 array's maximum read speed is a factor of the number of disks minus one, given that there is effectively one parity drive's worth of overhead.
- Usable capacity: (n − 1) × C.
- Fault tolerance: 1 drive.
- Minimum drives: 3.
RAID 6 — Double Parity (RAID 6 Calculator)
RAID 6 is RAID 5 with a second, independent parity block per stripe. Two drives' worth of capacity go to parity, and in exchange any two drives can fail simultaneously without losing data. That second parity is exactly what covers the dangerous window during a rebuild: if a second drive dies while the first is being reconstructed, the array survives.
This is the reason RAID 6 has largely replaced RAID 5 for arrays built from large drives. The cost is one more drive of overhead and a heavier write penalty, because two parity calculations have to be performed for every stripe.
- Usable capacity: (n − 2) × C.
- Fault tolerance: 2 drives, in any position.
- Minimum drives: 4.
RAID 1E — Striping and Mirroring (RAID 1E Calculator)
RAID 1E takes the mirroring idea of RAID 1 and the striping idea of RAID 0 and interleaves them across the whole array, so that every block exists twice, on two different drives. Its party trick is that it works with an odd number of drives — three, five, seven — which plain RAID 10 cannot do.
Half the raw capacity is usable, exactly as with RAID 10. One drive failure is always survivable; more than one is survivable only if the failed drives do not hold copies of the same blocks.
- Usable capacity: n × C ÷ 2.
- Fault tolerance: 1 drive guaranteed, up to ⌊n/2⌋ if you are lucky about which drives die.
- Minimum drives: 3.
RAID 10, 50 & 60 — Nested RAID Levels
Nested (or hybrid) RAID levels build one RAID array out of several smaller RAID arrays. The second digit is the level used inside each sub-array, and the first digit is the level used to join them together — always striping, RAID 0.
- RAID 10 (1+0) — a stripe across mirrored pairs. Half the raw capacity is usable, reads scale with every drive, and writes scale with half of them. It has the fastest rebuild of any redundant level, because restoring a failed drive is a straight copy from its mirror rather than a parity calculation across the whole array. It needs an even number of drives, minimum 4.
- RAID 50 (5+0) — several RAID 5 sub-arrays striped together. Usable capacity is (n − g) × C for g sub-arrays, and each sub-array can lose one drive. Minimum 6 drives, at least 3 per sub-array.
- RAID 60 (6+0) — several RAID 6 sub-arrays striped together. Usable capacity is (n − 2g) × C, and each sub-array can lose two drives. Minimum 8 drives, at least 4 per sub-array.
The point of RAID 50 and 60 is rebuild scope: a failure only forces a rebuild of the sub-array it happened in, not the whole array. With 24 drives, a RAID 60 of four sub-arrays rebuilds across 5 drives instead of 23.
RAID 5E and RAID 5EE
Both of these are RAID 5 with a hot spare folded into the array rather than sitting idle next to it. All the drives take part in every stripe, so you get the read performance of a wider array, and when a drive fails the rebuild starts immediately without anyone walking to the rack.
- RAID 5E — the spare capacity is reserved as a contiguous block at the end of every drive.
- RAID 5EE — the spare capacity is distributed through the array in the same way parity is, which makes the rebuild noticeably faster.
Both give usable capacity of (n − 2) × C: one drive's worth for parity, one drive's worth for the spare. Both tolerate one drive failure. Minimum 4 drives.
Which RAID Level Should I Choose?
- You need raw speed and the data lives elsewhere → RAID 0. Scratch space, game installs, render caches.
- Two drives, boot volume, small server → RAID 1. Simple, predictable, survives a drive death.
- Databases, virtual machines, anything with heavy random writes → RAID 10. No parity write penalty, and rebuilds are fast.
- Bulk storage on large drives → RAID 6. The second parity covers the rebuild window that kills RAID 5 arrays.
- Maximum usable space with some protection, on small or fast drives → RAID 5. Sensible for SSDs and drives under about 2 TB, where rebuilds are short.
- Very large arrays (16 drives and up) → RAID 50 or RAID 60, to keep each rebuild confined to one sub-array.
- An odd number of drives and you still want redundancy → RAID 1E.
- No one on site to swap a failed drive → RAID 5E or 5EE, so the rebuild starts on its own.
Set the optimize for selector in the calculator and it will rank the levels for you using the drive count you entered.
How to Use This RAID Calculator
- Choose your unit system: decimal (TB, as sold) or binary (TiB, as reported by the OS). Results are always shown in both.
- Pick a RAID level. The list shows the minimum drive count for each one.
- Enter the number of drives and the capacity of one drive. If your drives are different sizes, enter the smallest — RAID treats every drive as if it were the smallest one in the array.
- For RAID 50 or RAID 60, enter the number of sub-arrays. The drive count must divide evenly among them.
- Enter the sequential speed of a single drive in MB/s. Roughly 150–250 MB/s for a 7200 rpm hard drive, 500 MB/s for a SATA SSD, 3000 MB/s and up for NVMe.
- Optionally enter the price of one drive and your currency to get the total cost, the cost per usable terabyte, and how much of that money is going to redundancy rather than storage.
You get the usable capacity, the storage efficiency, the fault tolerance, read and write throughput, an estimated rebuild time, and a table comparing every RAID level using the same drives.
Worked Example: Comparing RAID 6 vs. RAID 10
Say you have 8 drives of 4 TB each, rated at 200 MB/s, costing $120 apiece. That is 32 TB raw and $960 spent, whichever level you pick.
- RAID 6: usable = (8 − 2) × 4 = 24 TB (21.83 TiB), 75% efficiency. Two drives may fail anywhere in the array. Read and write both scale ×6 → 1.20 GB/s. Cost per usable TB: $960 ÷ 24 = $40.00.
- RAID 10: usable = 8 ÷ 2 × 4 = 16 TB (14.55 TiB), 50% efficiency. One drive may fail anywhere; up to four may fail if no two of them are a mirrored pair. Reads scale ×8 → 1.60 GB/s, writes ×4 → 800 MB/s. Cost per usable TB: $960 ÷ 16 = $60.00.
The verdict. RAID 6 gives you 50% more space for the same money and a stronger guaranteed fault tolerance — two drives, no matter which ones. RAID 10 gives you faster writes and, more importantly, a rebuild that is a plain drive-to-drive copy instead of a parity reconstruction that reads every remaining drive. If this array is a file store or a backup target, take RAID 6. If it is running databases or virtual machines, take RAID 10 and pay the $20 per terabyte.
FAQs
Is RAID a backup?
No. RAID protects against a drive failing. It does not protect against deleting the wrong folder, ransomware, a failed controller, a power surge, theft or fire — all of which are written faithfully to every drive at once. Keep a real backup somewhere else.
Why is my 4 TB drive only 3.64 TB in Windows?
Because the two are counting in different systems. The manufacturer sells 4 × 10¹² bytes (4 TB decimal). Windows divides by 1024 three times and labels the result "TB", although it is really 3.64 TiB. No capacity is missing. Switch this calculator to the binary unit system to see the numbers your operating system will show you.
What happens if my drives are different sizes?
The array is limited by the smallest drive: every drive contributes only as much as the smallest one does, and the rest of the space on the larger drives is wasted. Enter the smallest capacity in the calculator.
How many drives can fail?
The calculator shows the guaranteed figure — how many drives can fail in any position — and, where it is higher, the best case in brackets. RAID 10 with 8 drives tolerates 1 drive guaranteed, but up to 4 if no two failures land on the same mirrored pair. Plan for the guaranteed number.
Why is RAID 5 discouraged on large drives?
Because of the rebuild window. Reconstructing a 16 TB drive means reading every remaining drive from end to end, which can take a day or more, and during that entire time a single further failure destroys the array. RAID 6 exists precisely to cover that window.
Are the speeds in this calculator realistic?
They are theoretical sequential maxima: the number of drives that can serve a stripe, multiplied by the speed of one drive. Real arrays are limited by the controller, the bus and the workload, and small random writes on RAID 5 and RAID 6 suffer a read-modify-write penalty that these multipliers do not model. Use them to compare levels against each other, not as a benchmark prediction.
How long does a rebuild take?
The estimate here is one drive's worth of data divided by the single-drive speed — the best case, on an idle array. On a controller that throttles rebuilds to keep serving users, expect it to take several times longer.