A backup repository is capacity-led, sequential and resilience-first — so the best RAID maximises usable space and survives failures during long retention. Size it in the RAID calculator, and remember the backup itself still needs an air-gapped copy.
What a backup target needs
Backup repositories take large, mostly sequential writes (nightly jobs) and occasional reads (restores). They're not IOPS-bound, so flash is rarely justified for the bulk — large nearline HDDs in a parity array give the most resilient capacity per pound. The priority is surviving drive failures over a long retention window without losing the backups.
That means dual parity: RAID 6 or RAID 60 for very large repositories, or RAIDZ2 on ZFS-based backup appliances. Single parity is too risky on the big drives backup repos use.
RAID is one layer, not the plan
The backup repository's RAID protects it from a drive failure — but a backup is supposed to be your recovery copy, so it must itself follow 3-2-1-1-0: multiple copies, different media, one off-site, one immutable or air-gapped, zero errors after verification. Ransomware specifically targets connected backups, so the air-gapped copy (object lock, or LTO tape on a shelf) is what survives an attack.
So: RAID 6/RAIDZ2 for the disk repository, plus an immutable/offline copy. Don't let a resilient RAID array lull you into a single-copy backup.
Sizing
Size for full backups plus the change rate across your retention period, with growth headroom — backup datasets compress and dedupe, so usable requirements depend on your software. Use the calculator for the array's usable capacity and resilience, and our backup & cyber-resilience page for the wider architecture.
For the air-gapped tier, LTO tape gives the lowest cost-per-TB and a true offline copy.