RAID expliqué pour serveurs domestiques

RAID combines multiple drives so they behave as one, improving performance and/or protecting against drive failure. Here's what each level does.

The common RAID levels

• RAID 0 (striping) — data split across drives for speed. No redundancy: lose one drive, lose everything.
• RAID 1 (mirroring) — an exact copy on two drives. Survives one drive failure; you get half the raw capacity.
• RAID 5 (single parity) — needs 3+ drives; survives one failure; loses one drive of capacity to parity.
• RAID 6 (double parity) — needs 4+ drives; survives two failures; loses two drives of capacity.
• RAID 10 (mirrored stripes) — needs 4+ drives; fast and resilient; half the capacity.

Which should you choose?

• Two drives, simple mirror → RAID 1.
• Four+ large drives, capacity-focused → RAID 6 (safer during rebuilds than RAID 5).
• Performance for VMs → RAID 10.

Use our RAID calculator to see usable capacity and fault tolerance for any configuration.

The most important point

RAID is not a backup.

RAID protects against hardware failure and improves uptime. It does not protect against:

• Accidental deletion
• Ransomware or corruption
• Fire, theft or power surges

All of those can destroy every copy in the array at once. You still need real backups following the 3-2-1 rule: three copies, on two media, with one offsite.

What about ZFS?

Modern filesystems like ZFS provide RAID-like redundancy plus checksums that detect and repair silent corruption (bit rot). For a serious NAS, ZFS is often a better choice than traditional hardware RAID.

RAID keeps your server running through a drive failure — but pair it with backups to actually keep your data safe.