A “backup completed” message tells you the copy operation has finished. To know whether you could resume work after a failure, you need to check which files you can restore, how recent they are, which tools you need and how long recovery takes.

How many copies do you actually need?

The 3-2-1 rule is a starting point, not a commandment: three copies, including the original and two backups, on two types of media, with one copy off-site. It covers media failure, human error and physical damage, which are three different risks that do not exclude each other.

One risk to account for is propagated deletion. Synchronisation alone does not guarantee a backup: a deletion can propagate to the other connected devices. You need previous versions or separate copies, not just synchronisation. And the medium matters less than the method: the speed of an SSD does not establish whether its backup copy is isolated, intact or recoverable.

What separates a copy from a backup

PropertySynced folderBackup
Recovery after deletionif the service retains recoverable versions or deleted itemsif valid versions exist and restoration is verified
Isolation from ransomwarenot guaranteed by synchronisation aloneoffline/immutable copies and separate access reduce risk
Integrity verificationdepends on the service’s featuresneeds to be planned and checked
Access during recoverydepends on where the copy residesmay be local or remote

The NCSC recommends verifying backups with real restore tests and keeping a recent offline copy ( UK NCSC ). These are goals to adapt to your data: an isolated copy still needs to be readable, decryptable and available when you need it.

How to test a restore

  1. Pick a representative set, not everything: a project, a photo folder, a database.
  2. Restore into a new destination, never over the originals.
  3. Open the files with the real tools, not just check that they exist. For a project, verify linked assets and configuration too.
  4. Time it. Time is part of the result: a three-day restore may be unsuitable when work must resume within hours.
  5. Record date, outcome, duration and problems. Include missing files, linked resources and unavailable credentials in the record.

Repeat the test at regular intervals and after significant changes to the backup or recovery setup.

When something is missing

  • Can the encryption key be recovered without the computer you lost?
  • How many versions does the copy keep, and for how long?
  • Would an authorised person other than you understand the procedure?
  • How much work would be lost between the last valid backup and the failure?
  • If ransomware encrypted the main drive, could credentials available to that same account also be used to access and modify the backup?

Restore to a separate destination without overwriting the originals. Record missing files, elapsed time and any credentials you needed. Those results give you a practical basis for choosing backup frequency and how long to keep older versions.

Sources

Report a correction ↗