RAID 5 Calculator

RAID 5 writes one parity block into every stripe and rotates which drive holds it from one stripe to the next, so the parity is distributed across all of the drives instead of sitting on a dedicated one. Added up, it takes one drive’s worth of capacity however many drives are in the array. Usable capacity is the number of drives minus one, multiplied by the capacity of the smallest drive. The minimum is three drives, and the calculator will not go below that.

As a worked example, the calculator above opens on four drives at 8 TB each. It reports 32.0 TB of raw capacity, 24.0 TB usable, and 75.0% efficiency: three drives’ worth of space holds data and one drive’s worth holds parity. The drive layout marks a single drive as parity to show how much space parity takes; on a real array those parity blocks are spread across every drive. Capacities are decimal terabytes, where 1 TB is 1,000 GB, which is how drives are labeled when you buy them.

The calculator assumes identical drives. An array is built on its smallest member, so when yours differ, enter the smallest size.

A RAID 5 array keeps running through the failure of any one drive. While that drive is missing the array is degraded: every read that touches it is reconstructed from the data and parity on the other drives, so it is slower, but all of the data is still readable. Once a replacement drive is in place, the controller rebuilds the missing contents from the survivors.

There is no second margin. If another drive fails before the rebuild completes, each stripe is missing two blocks, and a single parity block can only reconstruct one of them, so the array is lost. The fault tolerance of 1 that the calculator reports means one drive, whichever drive it is, and no more.

Reads stripe across the drives, so large sequential reads scale with the size of the array. Small random writes cost more. The controller cannot simply overwrite one block: it reads the old data block and the old parity block, computes the new parity, and writes the new data block and the new parity block back. That is four disk operations for one logical write, which is why RAID 5 is described as having a write penalty of four. RAID 6 pays six for the same write because it has two parity blocks to update, so RAID 5 is the lighter of the two for write-heavy work, though neither is a good fit for constant small random writes.

RAID 5 suits smaller arrays built from smaller drives, read-heavy work such as file shares and archives, and arrays that have backups behind them. Of the levels with redundancy, it gives the most usable capacity from a given set of drives, because it gives up only one of them to parity.

Large drives are where it gets uncomfortable. To rebuild a failed drive, the controller reads every block on every surviving drive, and the bigger the drives, the longer that takes. If one of those reads hits an unrecoverable read error, there is no second parity to fall back on, so that stripe cannot be reconstructed. The RAID 6 calculator shows the same drives with a second parity block, which covers exactly that case at the cost of one more drive of capacity. The rebuild time and risk calculator estimates how long that window lasts for your drives and how exposed it is. To compare every level for the same drive count and size, use the general RAID calculator.