Comparison
Disaster recoveryvsRestore testing
Disaster recovery
the plan for the day a whole region, account or dataset is gone, written down and hopefully rehearsed once a year.
The prepared response to losing an entire environment — a region, an account, a dataset — as opposed to the ordinary failures your architecture absorbs. It is defined by two numbers, how much time and how much data you can lose, and everything else follows from them. Its recurring failure is that the plan exists and has never been executed, so nobody knows that the restore takes eleven hours or that the credentials to run it are in the environment that is gone.
Full entry →Restore testing
you restore the backup into a scratch environment on a schedule, because a backup nobody has restored is a hypothesis.
Regularly performing a real restore from backups and verifying the result, rather than trusting that the backup job exited zero. It routinely uncovers backups of the wrong database, encrypted backups whose key is gone, restores that take far longer than the objective, and permissions that only exist in the environment being replaced. It is the single highest-value reliability exercise available and the most commonly skipped.
Full entry →