Last updated: 2026-06-23

RAID Capacity

RAID — Redundant Array of Independent Disks — pools multiple physical drives into a single logical volume. The level you choose determines whether the array prioritises raw capacity, fault tolerance, or read/write speed, and how much raw storage is consumed by redundancy.

How each level uses disk space

RAID 0 stripes data across all drives without any parity or mirroring. Every byte of raw capacity is available for user data, and reads and writes are distributed across all disks simultaneously. The trade-off is total exposure: one failed drive destroys the entire array.

RAID 1 keeps a perfect mirror of every disk. If you have two 4 TB drives, you get 4 TB of usable space — the second drive holds an identical copy. Any one drive (or all but one) can fail without data loss.

RAID 5 requires at least three drives and stores one parity block per stripe, rotated across all drives. With $n$ drives of capacity $s$:

$U_5 = (n - 1) \cdot s$

Four 4 TB drives in RAID 5 yields 12 TB usable and tolerates one simultaneous failure.

RAID 6 adds a second independent parity block, so two drives can fail at the same time. With $n$ drives of capacity $s$:

$U_6 = (n - 2) \cdot s$

The same four 4 TB drives yield 8 TB usable but survive any two concurrent failures — important for large drives that can take many hours to rebuild.

RAID 10 mirrors pairs of drives and then stripes across the mirror pairs. Usable capacity is half the total raw:

$U_{10} = \frac{n}{2} \cdot s$

An even number of drives (at least four) is required. RAID 10 combines mirroring's fault tolerance with striping's read performance, making it the standard choice for latency-sensitive database workloads.

Redundancy overhead

The overhead percentage measures how much raw capacity is dedicated to protection rather than user data:

$\text{overhead} = 1 - \frac{U}{n \cdot s}$

RAID 0 has 0% overhead; RAID 1 has 50%; RAID 5 with four drives has 25%; RAID 6 with four drives has 50%; RAID 10 always has 50%.

Worked example — RAID 5 with six 8 TB drives

Raw capacity: $6 \times 8\ \text{TB} = 48\ \text{TB}$

Usable (RAID 5): $(6 - 1) \times 8\ \text{TB} = 40\ \text{TB}$

Overhead: $1 - 40/48 \approx 16.7\%$

Fault tolerance: 1 disk.

If a drive fails, the array enters a degraded state and the controller rebuilds the lost data from parity onto a hot spare. During rebuild, performance drops and a second failure would be catastrophic — a key reason to consider RAID 6 for large arrays.

Choosing a level

Use RAID 0 only when you need maximum throughput and keep independent backups elsewhere. Choose RAID 1 for small two-drive setups such as a boot volume. RAID 5 suits three- to eight-drive NAS arrays where capacity efficiency matters. Move to RAID 6 once individual drives exceed 2 TB or the array has six or more members. RAID 10 is the enterprise default when both performance and redundancy are non-negotiable.

Uptime SLA Calculator pairs well here if you need to translate drive reliability into expected downtime budgets.