Disaster Recovery Processes

Disaster recovery for the example service covers the loss of an entire environment, not just a single failed server. The process assumes the primary account is unavailable and describes how to bring the service back up from backups in a separate account.

The recovery process has three phases: assess the scope of the loss, restore data from the most recent verified backup, and redirect traffic to the recovered environment once it passes the same health checks used for a normal deployment.

The most important number in a disaster recovery plan is not how fast you can recover — it is how recently you last tested the plan.

Disaster recovery drills are run on a regular schedule, with results and any gaps found recorded so the process keeps improving over time.