RAID Calculator

Using the RAID calculator

Fill in five inputs and press Calculate:

• RAID Type: the level you are considering, from RAID 0 to RAID 60.
• Drive Capacity: the size of one drive in TB, GB, TiB or GiB. TB and GB are decimal units, TiB and GiB are binary units (see the section on units below).
• Drive Cost: the price of one drive in any currency. Leave it empty or at 0 to hide the cost figures.
• Drives per RAID: how many drives are in one RAID group.
• RAID Groups: how many identical groups there are. For most levels this simply multiplies the array; RAID 50 and RAID 60 need at least two.

The results show total usable storage, cost per usable unit and total cost, disk space efficiency, the usable capacity of one group, fault tolerance per group, the total number of drives and, for levels where the figure is meaningful, the read and write speed gain.

If the numbers do not fit the chosen level, you get a message instead of a result. RAID 1 accepts exactly 2 drives; RAID 1E, 4 and 5 need at least 3; RAID 6 needs at least 4; RAID 10 needs at least 4 and an even number; RAID 50 needs at least 6 drives in each group; RAID 60 needs groups of at least 4. RAID 50 and 60 also need two or more groups.

What RAID does

RAID stands for Redundant Array of Independent Disks. It combines several physical drives into one logical volume, trading some capacity, cost or speed for protection or performance. The idea was laid out in a 1988 paper by Patterson, Gibson and Katz, who showed that arrays of small inexpensive drives could rival one large expensive one.

Three techniques are mixed and matched by the different levels:

• Striping spreads consecutive blocks of data across drives so that they can be read or written in parallel. It boosts speed but adds no protection.
• Mirroring stores the same data on more than one drive. A mirror survives a drive failure and can serve reads from any copy, but half the raw space goes to the copy.
• Parity stores a computed check value (an XOR of the data blocks) so that the contents of a lost drive can be rebuilt from the survivors. It costs one or two drives' worth of space rather than half the array.

Capacity and fault tolerance by RAID level

In the table, n is the number of drives in one group and c the capacity of one drive. Usable capacity is per group; the calculator multiplies by the number of groups.

RAID 10 can survive up to one failure per mirrored pair, which is where the calculator's range of "1 to n/2" comes from. The guaranteed figure is still one, because a second failure in the same pair destroys the data.

For RAID 10 the headline total follows the half-of-all-drives rule above. The separate "single RAID group" line uses a different convention, (n − 2) × c, and so does not add up to the total for RAID 10. Rely on the total when planning.

How capacity, efficiency and cost are worked out

efficiency = usable capacity ÷ (drives × drive size) × 100

cost per unit = (drives × drive price) ÷ usable capacity

Example 1, RAID 5. Four 4 TB drives at $100 each. Usable capacity is (4 − 1) × 4 = 12 TB out of 16 TB raw, so efficiency is 75%. The array costs 4 × $100 = $400, which is $400 ÷ 12 = $33.33 per usable terabyte.

Example 2, RAID 6. Six 4 TB drives: (6 − 2) × 4 = 16 TB usable of 24 TB, or 66.67%. The same six drives in RAID 10 would give 6 × 4 ÷ 2 = 12 TB, or 50%.

Example 3, RAID 60. Two groups of five 4 TB drives: each group holds (5 − 2) × 4 = 12 TB, so the array offers 24 TB of 40 TB, or 60%.

Parity levels get more efficient as you add drives, because the parity cost stays at one or two drives. The trade-off is a bigger array to rebuild and more drives that can fail.

Read and write speed gain

For RAID 0, 1, 5, 6 and 10 the calculator shows an idealised speed relative to a single drive. Reads can in theory use every drive, so the read gain equals the number of drives. Writes cost extra work: a mirror writes each block twice, RAID 5 reads and rewrites data and parity (about four disk operations per small write), and RAID 6 does so for two parities (about six). The write gain is therefore the drive count divided by 1 for RAID 0, 2 for mirrors, 4 for RAID 5 and 6 for RAID 6, rounded to a whole number.

These are rules of thumb for small random writes; caches, SSDs and software implementations change the real picture. No speed rows are shown for RAID 1E, 4, 5E, 50 and 60.

Drive size units: TB versus TiB, GB versus GiB

Drive makers sell capacity in decimal units: 1 TB is 10¹² bytes. Operating systems such as Windows report binary units, where 1 TiB is 2⁴⁰ = 1,099,511,627,776 bytes (and often print it as "TB"). The binary unit is about 10% bigger at the terabyte level, so a 4 TB drive shows up as roughly 3.6 TiB.

When you choose TB or GB, the results show the decimal figure first and the binary figure in brackets, for example 12 TB (10.91 TiB). The bracketed number divides by the rounded factors 1.1 for terabytes and 1.074 for gigabytes, so it can differ from the exact binary value by a few hundredths of a percent. If you choose TiB or GiB, results stay in those units. Use the TB to GB converter or GB to TB converter for other conversions.

RAID is not a backup

A redundant array keeps your data available when a drive dies. It does not protect against accidental deletion, malware or ransomware, file system corruption, a failed controller, theft, fire or flood, because every change is faithfully copied across the array. Keep separate backups on other media or in another location.

Large drives add a further risk. Rebuilding a failed drive reads every other drive in the group, which can take many hours or days, and a second failure or unreadable sector in that window can lose the array. That is why double parity (RAID 6) or mirrors are popular for big drives.

Once your storage is sized, the download time calculator can estimate how long it will take to fill it over your connection.