RAID stands for Redundant Array of Independent (originally Inexpensive) Disks: combining multiple drives into one logical volume for redundancy, performance, or both. The concept dates to the late 1980s, when disks meant hard drives; today RAID applies equally to arrays of SSDs, and the trade-offs have shifted as drives have grown to sizes the original designers never imagined. Use our RAID Calculator to see exactly how much usable capacity each level yields from your drives.
The RAID levels that matter
RAID 0 (striping)
Data split across all drives with no redundancy. Full capacity, best speed, and any single drive failure loses everything. Appropriate for scratch space and nothing you cannot re-create.
RAID 1 (mirroring)
Every write goes to two or more drives, so the array survives a drive failure with no data loss and reads can be served from either copy. You pay half your raw capacity for that safety. Two-bay NAS units are the classic RAID 1 home, and it remains the simplest, most predictable redundancy there is.
RAID 5 (striping with parity)
Data plus one drive’s worth of parity spread across the set; survives one drive failure while giving up only one drive of capacity. The catch in 2026: rebuilding after a failure means reading every remaining drive end to end, and with 30TB-class drives that window stretches to a day or more of degraded, vulnerable operation.
RAID 6 (dual parity)
Like RAID 5 but survives two simultaneous failures, at the cost of two drives of capacity. As rebuild windows have grown with drive capacity, RAID 6 (or a second parity drive generally) has become the sane default for large arrays built on big disks.
RAID 10 (mirrored stripes)
Stripes across mirrored pairs: strong performance, fast rebuilds (only the failed mirror is copied), half your raw capacity. The performance-and-safety pick when capacity efficiency matters less.
What changed: capacity outgrew the rebuild
RAID math was designed around drives measured in gigabytes. With NAS drives now at 32TB and enterprise HAMR models like the Exos M 30TB sustaining roughly 275MB/s, a full rebuild of one drive takes on the order of a day under ideal conditions, longer under load. That window is when a second failure hurts, which is why large home and SMB arrays increasingly run RAID 6, why enterprise systems moved to declustered and erasure-coded schemes, and why hyperscale storage abandoned classic RAID for object storage entirely. RAID is not obsolete; it just needs to be chosen with rebuild time in mind, not only usable capacity.
RAID is not a backup
RAID protects against drive failure, not deletion, ransomware, fire, or theft. An array and a backup are different layers; you need both. This is also where recertified drives earn their place: in a redundant array with backups behind it, recertified enterprise drives cut the cost of capacity substantially at low practical risk.
Choosing in practice
For a two-bay NAS: RAID 1. Four to eight bays of large drives: RAID 6 or the vendor’s dual-redundancy equivalent, with drives from our Leaderboard’s NAS picks. Performance arrays of SSDs: RAID 10 or the platform’s native equivalent. A practical example of multi-bay RAID done well is the Ubiquiti UNAS Pro 8 we reviewed. Then run the numbers in the RAID Calculator before you buy anything.
Amazon