RAID Levels Explained: RAID 0, 1, 5, 6 and 10
Basics4 minInnovialabs editorial team
In short
RAID combines several drives into one array for speed or to survive a drive failure: RAID 1 and 10 mirror, RAID 5 and 6 use parity, RAID 0 only adds speed — and no level replaces a backup.
The short answer
RAID combines several physical drives into one logical volume. Depending on the level, the array gains speed, resilience (it survives one or two failed drives) or both, at the cost of some usable capacity.
Three key ideas:
- Striping — data is split into blocks and written across several drives in parallel.
- Mirroring — the same data is written to two drives.
- Parity — check data that lets the array reconstruct the contents of a failed drive.
Levels compared
N is the number of equal-size drives.
| Level | Min. drives | Usable capacity | Survives failure of | Speed |
|---|---|---|---|---|
| RAID 0 | 2 | N drives | None | Fastest reads and writes |
| RAID 1 | 2 | 1 drive | All but one | Good reads, writes like a single drive |
| RAID 5 | 3 | N − 1 | 1 drive | Good reads, slower writes due to parity |
| RAID 6 | 4 | N − 2 | Any 2 drives | Good reads, writes slower than RAID 5 |
| RAID 10 | 4 | N / 2 | One per mirrored pair | Fast reads and writes |
Each level in plain words
RAID 0 — striping with no protection. Capacity and speed add up, but losing any drive destroys the whole array. Only for temporary data you can afford to lose.
RAID 1 — a mirror. Two drives hold identical data. The simplest, most predictable option for a small two-drive NAS or server.
RAID 5 — striping with distributed parity. You lose one drive's worth of capacity, and the array survives one failed drive. Cost-efficient, but it has a serious weakness with large drives (see below).
RAID 6 — like RAID 5 with double parity. You lose two drives' worth of capacity, but the array survives any two failures. A good choice for arrays of many large drives.
RAID 10 — mirrors joined by striping. Half the capacity goes to redundancy, but writes are fast and rebuilds are simpler: only the mirrored pair is copied. Often chosen for databases and virtual machines.
Rebuild risks on large drives
When a drive fails, the array runs in degraded mode; after you replace the drive, a rebuild starts. This is the most dangerous moment:
- Rebuilding large drives can take many hours or even days, and during that time the array has no redundancy left (RAID 5) or less of it.
- A rebuild reads every remaining drive in full. If one of them hits an unreadable sector, RAID 5 may fail to recover part of the data or the whole array.
- Drives from the same batch and of the same age tend to wear out together, so a second failure can follow the first.
- The rebuild puts extra load on the drives, and performance for users drops.
That is why RAID 6 or RAID 10 is generally preferred over RAID 5 for large drives.
Why RAID is not a backup
RAID only protects against physical drive failure. It will instantly and faithfully replicate across all drives:
- accidental deletion or overwriting of a file;
- files encrypted by ransomware;
- file system corruption or a controller fault.
It does not help with fire, theft or flooding either, since all drives sit in the same box. A backup is a separate copy of the data in another place, ideally with version history.
How to choose
- 2 drives → RAID 1.
- 3–4 drives, file archive, budget matters most → RAID 5 on smaller drives, or RAID 6.
- Many large drives → RAID 6.
- Databases, virtual machines, write speed matters → RAID 10.
- Temporary data, speed only → RAID 0.
Also: turn on drive health alerts, keep a spare drive on hand and monitor SMART data.