Case study
Building recovery before it is needed
The problem
The infrastructure had backups. It had neither a written restore procedure, nor proof that a restore worked, nor an estimate of the time needed to bring a service back.
Application workloads on Kubernetes were not backed up at all, a gap that the move from Azure Web Apps to AKS had made critical without the backup setup keeping up.
The solution
AKS cluster backups set up with Velero, covering Kubernetes resources and persistent volumes.
Disaster recovery plan designed and documented: failover scenarios, service recovery order, and a runbook that someone who did not build the system can follow.
Restores tested, not just scheduled. This is the part most setups leave out, and the one that makes the difference on the day it is needed.
Hover over or select a step to read the details.
The results
- Cluster backup and restore in working order
- A documented recovery plan that a third party can execute
- Failover scenarios proven by real tests
What I take away from it
A recovery setup is only worth its last successful restore. A green backup report proves nothing: it proves that a file was written, not that a service can restart.