RAID 10 Calculator

RAID 10 groups the drives into pairs, mirrors each pair so that both drives hold the same data, and stripes data across the pairs. Every block is stored twice, so half of the raw capacity is usable however large the array is. Usable capacity is the number of drives divided by two, multiplied by the capacity of the smallest drive. The minimum is four drives, which makes two mirror pairs, and the count needs to be even so that every drive has a partner. The calculator does not round an odd count down, so enter an even number.

As a worked example, the calculator above opens on eight drives at 8 TB each. It reports 64.0 TB of raw capacity, 32.0 TB usable, and 50.0% efficiency: four mirror pairs, each contributing one drive’s worth of space. Unlike the parity levels, the efficiency does not improve as drives are added; it stays at 50%. 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. Each pair is limited to its smaller drive and the stripe to its smallest pair, so when yours differ, enter the smallest size.

The guaranteed figure is one drive. Any single drive can fail, and its mirror partner still holds a complete copy of everything that was on it.

Beyond one failure, the outcome depends on which drives fail. The array keeps running as long as every mirror pair still has one working drive, so with eight drives it can lose as many as four, provided each comes from a different pair. If both drives of the same pair fail, the data on that pair has no other copy, and because data is striped across all of the pairs, the whole array is lost. A second failure is survivable or fatal depending only on where it lands. The comparison table shows this as “1 (up to 4)”, and the Fault Tolerance card gives the guaranteed figure with the best case beneath it. For planning, count on the guaranteed figure.

RAID 10 has no parity to compute. A small write goes to both drives in its pair, which is two disk operations for one logical write, a write penalty of two, compared with four for RAID 5 and six for RAID 6. Reads can be served by either drive of a pair.

Rebuilds are simpler too. Replacing a failed drive is a straight copy from its surviving mirror partner, so the rebuild reads one drive instead of the whole array, and the other pairs carry on serving data without taking part. Less data is read, so the rebuild finishes sooner and less of the array is exposed while it runs.

RAID 10 fits write-heavy workloads such as databases and virtual machine storage, where small random writes dominate and where getting back to full redundancy quickly after a failure matters more than squeezing out capacity.

The price is that capacity. Half of the raw space always goes to the mirrors, so the gap against the parity levels widens as the array grows: for the same eight drives, the comparison table shows RAID 6 at 75.0% efficiency against 50.0% for RAID 10. When the workload is mostly reads and capacity matters more, the RAID 6 calculator covers the capacity-focused alternative, which survives any two failures. To compare every level for the same drive count and size, use the general RAID calculator.