RAID 6 Calculator

RAID 6 writes two independent parity blocks into every stripe, so two drives’ worth of capacity go to parity however many drives are in the array. Usable capacity is the number of drives minus two, multiplied by the capacity of the smallest drive. The minimum is four drives, and the calculator will not go below that.

As a worked example, choose eight drives at 8 TB each. The calculator reports 64.0 TB of raw capacity, 48.0 TB usable, and 75.0% efficiency: six drives’ worth of space holds data and two drives’ worth holds parity. Capacities are decimal terabytes, where 1 TB is 1,000 GB, which is how drives are labeled when you buy them.

Mixing drive sizes wastes the difference. An array is built on its smallest member, so a 12 TB drive alongside 8 TB drives contributes 8 TB and the other 4 TB sit idle until every drive has been replaced with a larger one. The calculator assumes identical drives, so when yours differ, enter the smallest size.

A RAID 6 array keeps running through any two drive failures at once, whichever two drives they are. With one or two drives failed the array is degraded: reads that touch a missing drive are reconstructed from the surviving data and parity, so it is slower, but all of the data is still readable and the array can be rebuilt once replacement drives are in place.

The protection has a hard edge. If a third drive fails before the rebuild completes, the array is lost, because the stripe no longer holds enough information to reconstruct the missing blocks. That is why the length of the rebuild window matters.

Reads stripe across the data drives, so large sequential reads scale with the number of drives in the array. Writes are where the cost shows. A small random write cannot simply overwrite one block: the controller reads the old data block and both old parity blocks, computes the new parity, and writes the new data block and both new parity blocks back. That is six disk operations for one logical write, which is why RAID 6 is described as having a write penalty of six.

The practical consequence is that RAID 6 suits read-heavy and capacity-focused workloads, such as file servers, media libraries, and backup targets, better than it suits write-heavy databases with constant small writes.

RAID 5 spends one drive on parity instead of two, so it offers more usable capacity from the same drives, but it survives only a single failure. Large drives mean long rebuilds, and during that window RAID 5 has no margin at all: a second drive failure, or an unrecoverable read error on a surviving drive while the rebuild reads every block of it, means data loss. RAID 6 puts a second drive of margin into exactly that window.

RAID 10 takes the other route. It mirrors pairs of drives and stripes across the pairs, so a rebuild is a straight copy from the other half of one mirror, and small writes carry no parity penalty. The price is capacity: half of the raw space goes to the mirrors, where RAID 6 gives up two drives however large the array, so the gap widens as the array grows. If capacity and two-drive tolerance matter more than write speed, RAID 6 is usually the choice. If write performance and fast rebuilds matter more, RAID 10 is. The general RAID calculator puts the two side by side for the same drive count and size.