RAID is not a backup. Mirrors and parity protect against a drive failing, not against ransomware, accidental deletion or a fire. Keep a separate copy of anything important. Capacity figures here use simplified assumptions: identical drives, decimal sizes (1 TB = 1,000 GB), no hot spare, and no controller reserved space.
One drive worth of distributed parity is spread across the array. That parity replaces the capacity of exactly one drive, which is the parity overhead.
Decimal TB, exactly as printed on the drive.
Usable capacity
28,000 GB (28 TB)
Raw capacity
32,000 GB (32 TB)
Overhead
4,000 GB (4 TB) (12.5%)
Failures tolerated
1
| Level | Formula | Usable now | Min drives | Failures tolerated |
|---|---|---|---|---|
| RAID 0 (striping) | N × S | 32,000 GB (32 TB) | 2 | 0 — one drive failure loses everything |
| RAID 1 (mirroring) | 1 × S | 4,000 GB (4 TB) | 2 | 7 (any drive except one in the mirror set) |
| RAID 5 (single parity) | (N − 1) × S | 28,000 GB (28 TB) | 3 | 1 |
| RAID 6 (double parity) | (N − 2) × S | 24,000 GB (24 TB) | 4 | 2 |
| RAID 10 (striped mirrors) | (N ÷ 2) × S | 16,000 GB (16 TB) | 4 | 1 guaranteed; up to 4 if each failure is in a different mirror pair |
Where the capacity goes. RAID 0 stripes with no redundancy, so all N drives are usable. RAID 1 mirrors, so only one drive worth is usable however many copies you keep. RAID 5 stores one drive worth of distributed parity, giving (N − 1) × S; RAID 6 stores two independent parity blocks, giving (N − 2) × S. RAID 10 mirrors pairs first, so half the raw capacity is always spent on copies.
Formula: usable capacity = usable drives × drive size, where N is the drive count and S is one drive. Everything is calculated in your browser from the numbers you type — nothing is stored or sent.