RAID is not a backup

expression

ops cultureIT support

The storage truth that redundancy protects against disk failure, not against deletion, corruption, or ransomware.

RAID keeps a volume alive when a drive dies - and faithfully mirrors every mistake to all disks at once. Deleted the file? It is gone, redundantly. Availability and recoverability are different problems; a backup solves the second one.

RAID protects against a disk failing. It does not protect against anything else, and the list of anything else is where all the data loss actually happens: deletion, corruption, ransomware, a botched migration, a controller writing garbage, fire, theft, and someone typing a command against the wrong host.

The confusion is understandable, because RAID does produce a redundant copy and redundancy sounds like backup. The difference is that a mirror is synchronous: whatever happens to the primary happens to the copy immediately, including deletion. A backup is a point-in-time copy that is deliberately not synchronized, which is precisely the property that lets you go back to before the mistake.

The same reasoning applies more broadly than people apply it. Cloud storage replication is not a backup for the same reason, since a deleted object is deleted in every replica, and a synchronized folder propagates encryption by ransomware to every device within minutes. The question that separates the two categories is simple: if I destroy the data right now, deliberately, does this mechanism still have it? If the answer is no, it is availability, not backup.

All glossary entries