DNSfinder.net
Storage Planning

RAID Calculator

Pick a RAID level, enter how many disks you have and the size of each, and the calculator shows the usable capacity, how much space goes to mirroring or parity, how many drives can fail before data is lost, and a rough read and write speed relative to one disk. It covers RAID 0, 1, 5, 6, 10, 50, 60 and JBOD, and is meant for anyone sizing a NAS or server array before buying drives.

Loading RAID Calculator...

Querying live edge resolvers and network infrastructure...

How to use the RAID Calculator

  1. Choose a level from the RAID level list. The calculator opens on RAID 5 with four 4 TB disks.
  2. Enter the number of disks (1 to 128) and the size of one disk, and pick GB or TB as the unit.
  3. For RAID 50 or RAID 60, set Spans to the number of RAID 5 or RAID 6 groups the disks are divided into.
  4. Read the four tiles and the capacity bar. If the disk count does not suit the level, a message states the requirement instead.

What to look for

Usable space in two units

The main figure uses the decimal units printed on drive labels. The line under it converts to binary units (TiB or GiB), which is what an operating system reports. Twelve usable terabytes show up as about 10.91 TiB, so nothing is missing when the OS shows a smaller number.

Fault tolerance is a worst case

RAID 5 reads 1 disk and RAID 6 reads 2 disks, both guaranteed. RAID 10 reads 1 disk per mirror pair: it can lose one drive from every pair, but losing both halves of a single pair ends the array. RAID 0 and JBOD show None, and the tile turns red.

Speed figures are multipliers

Read and write speeds are shown relative to a single disk and come from a simple model: reads scale with the data disks, RAID 5 writes are divided by four and RAID 6 writes by six to reflect the parity write penalty. Real results depend on the controller, its cache and the workload.

Guide

How RAID levels trade capacity for redundancy

What each level does with your disks

RAID 0 stripes data across all disks: you keep every byte of capacity and lose everything when one drive fails. RAID 1 mirrors, so any number of disks hold one disk's worth of data and the array survives until the last copy dies. JBOD, or spanning, joins disks end to end with no striping and no protection. The calculator reports the full raw capacity and no speed gain.

RAID 5 spreads one disk's worth of parity across the set, so n disks give n minus 1 of capacity and any single drive can fail. RAID 6 keeps two independent parity blocks, costs two disks and survives any two failures. RAID 10 stripes across mirrored pairs and always yields half the raw space. With the default four 4 TB disks, RAID 5 gives 12 TB, while RAID 6 and RAID 10 both give 8 TB.

  • RAID 0 and RAID 1: at least 2 disks.
  • RAID 5: at least 3 disks. RAID 6: at least 4.
  • RAID 10: an even number of disks, 4 or more.
  • RAID 50: at least 3 disks in each span. RAID 60: at least 4 in each span.

Nested levels and the Spans field

RAID 50 is a stripe across several RAID 5 groups, and RAID 60 a stripe across several RAID 6 groups. The Spans field sets how many groups there are, and the disk count must divide evenly between them. Each span gives up one disk to parity in RAID 50 and two in RAID 60, so twelve 4 TB disks in two spans yield 40 TB as RAID 50 and 32 TB as RAID 60.

The tolerance line for these levels reads 1 disk per span or 2 disks per span, with the best-case total in brackets. The per-span figure is the one you can count on: a RAID 50 array survives two failures only when they land in different spans. More spans mean smaller groups and shorter rebuilds, paid for in capacity.

Reading the result tiles

Usable capacity is what is left for data. Redundancy overhead is the space spent on mirrors or parity, also given as a percentage of the raw total. For four disks in RAID 5 it is 4 TB, 25% of raw 16 TB. The capacity bar draws the same split in green and amber, and labels the amber part mirror for RAID 1 and 10 and parity for the rest.

Speed vs one disk shows a read multiplier and an approximate write multiplier. Four disks in RAID 5 show 3.0 times read and 0.8 times write, because a small write to a parity array has to read the old data and old parity, then write both back. RAID 10 shows reads at the full disk count and writes at half of it, since mirrors pay no parity cost.

What the numbers leave out

The model assumes identical disks. Real arrays size every member to the smallest drive, so adding a 4 TB disk to a set of 8 TB disks wastes half of each larger one. Enter the smallest size to get a correct figure. Hot spares are not modelled either. Leave them out of the disk count, since a spare holds no data until it replaces a failed drive.

Filesystem and volume manager overhead comes on top. Formatting, metadata, snapshot reserves and the free space most filesystems need to stay fast all reduce what you can store, by an amount that depends on the system. Treat the usable figure as an upper bound.

Fault tolerance also says nothing about rebuild risk. Rebuilding a parity array reads every surviving disk from end to end, which takes many hours on large drives and is exactly when a second failure or an unreadable sector does the most damage. That is the usual argument for RAID 6 or RAID 10 over RAID 5 on big disks.

Knowledge Base & FAQ

RAID Calculator FAQs

Common questions about the raid calculator.

The capacity of all disks minus one. Four 4 TB drives give 12 TB and six give 20 TB, and the overhead falls as the set grows: 25% of raw space with four disks, about 17% with six. The calculator also shows the binary figure an operating system reports, about 10.91 TiB for 12 TB.