What RAID protects, and what still needs a backup

Begin with the failure you mean

People often use safe storage to describe several different outcomes. They may mean that a computer keeps working after a drive fails, that a deleted file can be recovered, or that data survives a stolen device. One storage feature rarely addresses all of those events.

RAID arrangements combine drives in ways that can improve performance, provide redundancy, or both, depending on the level. Some configurations can tolerate one or more drive failures; others provide no redundancy at all. The exact layout matters. Synology's drive-failure guidance distinguishes rebuilding a degraded RAID array from recovering after a volume crash, where a separate backup is essential.

That statement is not a criticism of RAID. It describes its role. A redundant array can help maintain access during a supported hardware failure while leaving many other causes of data loss untouched.

A mirror follows changes too

In a mirrored arrangement, the system maintains corresponding data across drives. That can be useful if one drive stops working. But when a user deletes a file through the filesystem, the array normally carries out that deletion as an intended change. The mirror is not an independent historical copy waiting to disagree with the mistake.

The same principle applies to unwanted edits or corrupted files written by an application. Redundancy preserves the storage operation; it does not determine whether the operation was desirable. Recoverable versions require an additional feature or backup process.

Snapshots can help with some mistakes when the system supports them, but their location and retention still matter. A snapshot on the same device can be lost with that device.

Usable capacity and rebuilds are tradeoffs

Redundancy consumes capacity and introduces management tasks. The usable space depends on drive sizes and the selected arrangement. A two-drive mirror, for example, does not provide the combined capacity of both drives for unique data. Other layouts have different calculations and requirements.

After a failed drive is replaced, the system may need to rebuild redundancy. During that period, performance and remaining fault tolerance can differ from normal operation. Follow the vendor's replacement procedure and compatibility guidance rather than pulling drives based on an assumption.

A rebuild can be a demanding operation, which is another reason to have a current independent backup before making storage changes. Do not treat a degraded array as a convenient moment to reorganize everything at once.

Assign protection by event

Use a short event map to see where the gaps are. The answers below describe common roles; the capabilities of a particular system still need to be verified.

Back up the array as a source

If the NAS or array holds the primary files, treat it as the source for a backup system. Select the important folders and include application configuration or catalogs where needed. Decide how often data changes and how much history must remain available.

Keep backup credentials separate from ordinary shared-folder access where the platform supports that design. A backup destination that every user can freely erase is less independent than it first appears. At home, a rotated disconnected copy can be one practical part of the plan, provided it is updated regularly.

Test restoration from the backup to a different location. Recover a file, an earlier version, and a representative application project. This checks the protection that RAID itself does not provide.

Use the terms accurately

When describing your setup, say which failures it can tolerate and where the backup lives. Two mirrored drives and a versioned offsite backup is more informative than three drives, so it is safe. Precise language makes missing pieces easier to notice.

Review the plan when adding capacity, changing RAID level, or moving to a new enclosure. Such changes can affect both usable space and the recovery procedure. Keep the existing backup intact throughout the transition.

RAID is useful when availability through certain drive failures matters. A backup is useful when you need to recover an earlier or separate copy. Using both deliberately provides a clearer plan than asking either one to do the other's job.