Instance disks, persistent volumes and snapshots
A cloud deployment can combine different storage types. An instance disk, an attached volume and a snapshot are separate resources with different lifecycle and recovery behaviour. Record where each important dataset lives before deploying or deleting anything.
Questions to settle before deployment
- Which disks are persistent, and what happens to them if the instance is rebuilt, resized, stopped or deleted?
- Can volumes be detached and attached to another instance, and what application shutdown or unmount procedure is required?
- How are snapshots created and retained, and are databases made consistent before the snapshot is taken?
- Is there an independent backup outside the immediate failure domain, and has restoration been tested?
Do not delete an instance until its required data and attached-resource behaviour are confirmed. Before detaching a live volume, coordinate an application-safe procedure. Snapshot availability alone does not establish a backup retention policy or a guaranteed recovery time.
