A backup is a copy of data kept so you can get it back after loss: a deleted file, a failed disk, ransomware, a fire. The copy is only worth what a restore can recover, so the real subject is recovery: how much you can lose, how long you can wait, and whether you have tried it.
Key concepts
- 3-2-1 rule: three copies, on two kinds of media, with one stored somewhere else.
- Full, incremental and differential: a whole copy, only what changed since the last backup, or what changed since the last full one.
- Restore testing: a backup that has never been restored is a hope, not a backup.
- Consistency: databases and running systems need snapshots or dumps so the copy is coherent.
- Protection of the copies: encrypt them and keep one out of reach of the systems it protects.
Vocabulary
- RPO: recovery point objective, the most data you can afford to lose, measured in time.
- RTO: recovery time objective, the longest you can afford to be down.
- Snapshot: a frozen, point-in-time view of a filesystem or volume.
- Retention: how long each backup is kept before it is deleted.
What to study, in order
- Copy a folder to another disk by hand, then script it with a tool such as rsync.
- Switch to a tool that deduplicates and encrypts, and keep several dated versions.
- Write down an RPO and an RTO for your own data and check that your schedule meets them.
- Back up a database with its own dump tool and restore it into an empty one.
- Keep one copy offsite or offline, and delete a file on purpose to restore it.
- Rehearse a full restore of a whole machine from nothing, and time it.