Backups · Reliability

A backup is not a recovery plan

28 Sep 2026 · 3 min read

Having backups sounds reassuring. But the important question is not whether a backup file exists. It is whether you can restore the website when something actually goes wrong.

A database can be corrupted, a deployment can break production, an account can be compromised, or a hosting problem can remove data. In those moments, an untested backup may be much less useful than expected.

Backups can fail silently

A scheduled backup job may stop running because of expired credentials, insufficient storage or a configuration change. If nobody checks it, the problem can remain unnoticed for weeks.

That is why backup monitoring matters. You should know when the last successful backup happened and whether the expected data was actually included.

The restore process matters just as much

A backup is only one half of the process. The other half is knowing how to restore the application, database and required files safely.

If restoration requires finding old credentials, guessing which file is correct or rebuilding infrastructure from memory, recovery becomes slower exactly when time matters most.

One copy is not enough

Keeping a backup on the same server as the website protects against some mistakes, but not against a full server failure or account-level incident.

A practical setup usually keeps backups separate from the production system and retains more than one recovery point, so a recent bad change does not overwrite the only usable copy.

Test recovery before you need it

You do not need a complicated disaster-recovery exercise for every small website. But periodically restoring a backup into a safe environment answers the most important question: can this data actually be used?

It also makes the recovery steps clearer, so an incident is not the first time anyone tries them.

The goal is predictable recovery

Backups are not about collecting files. They are about reducing the impact of a failure.

A good setup makes it clear what is backed up, how often it happens, where copies are stored, whether failures are detected and how the website can be restored.