Two numbers decide the design
How much data can you afford to lose — an hour, a day, a week? And how long can you afford to be unable to operate? Those two answers determine backup frequency and recovery approach, and they are business decisions rather than technical ones.
Most businesses have never been asked. Asking produces useful clarity, and frequently reveals that the current arrangement does not meet either number.
What people forget to back up
- Cloud services. Your email and file storage provider protects against their failure, not against your deletion. Retention windows are shorter than people assume.
- The website, including the database, not just the files
- Line-of-business systems hosted by a supplier — ask what their backup and restore commitment actually is
- Configuration, not just data: the settings that make everything work
- Anything on individual machines that exists nowhere else
The rule of three, still
Three copies, on two kinds of storage, one of them off-site and ideally offline or immutable. The immutable copy is what protects against ransomware, which will encrypt any backup it can reach.
If your backup is a drive plugged into the machine it backs up, you have a copy, not a backup. Ransomware encrypts both.
Test the restore, not the backup
A backup job reporting success proves it ran. It does not prove the data is complete, uncorrupted or restorable within your tolerance.
- Pick a realistic scenario: a deleted folder, a corrupted database, a lost machine
- Restore it somewhere safe, from the backup, without shortcuts
- Time it, and compare against your acceptable downtime
- Write down what you learned, because something always surprises you
Annually is enough for most businesses. Never is what most actually do.
Write down what happens in an incident
Who decides, who contacts whom, what gets restored first, and what the business does in the meantime. One page is enough and it is worth far more than an elaborate plan nobody has read.
Include contact details that do not depend on the systems being down, which is a detail routinely overlooked until it matters.