What is the difference between disk redundancy and backups?

They protect against different failures

Disk redundancy, such as a RAID mirror, means the data is written to more than one disk so the server can keep working if one disk fails. A backup is an independent copy taken at a point in time that you can restore from. The two answer different questions: "what if a disk dies?" and "what if the data itself is wrong or gone?"

Why a mirror is not a backup

A mirror copies everything immediately, including mistakes. If you delete a directory, an application bug corrupts a database, a bad deployment overwrites files or malware encrypts them, the second disk holds the same damage. Redundancy also usually lives in the same machine, so a fire, theft, controller fault or account problem can affect both disks. Not every RAID level is redundant either: striping without parity or mirroring improves capacity or speed but gives no protection from a failed disk.

What a useful backup looks like

• Stored away from the machine it protects, ideally with a different failure domain.
• Kept for long enough to go back before a problem began, not only the latest copy.
• Covering the database as well as files, using a method that produces a consistent copy.
• Restored on a schedule, so you know it works and how long recovery takes.

Questions to ask before renting a dedicated server

Ask what disk layout the server comes with and whether you can choose it. Ask who replaces a failed disk and how long that takes, and whether the provider offers backup storage or whether you must arrange your own. Decide how much data you can afford to lose and how long you can be offline; those two numbers should drive your choice of layout and backup frequency. Treat anything the provider has not documented as unverified.