Fifteen-minute recovery, nightly backups
A workload is contractually held to recovering within fifteen minutes and losing no more than a minute of data. The plan is nightly snapshots copied to a second region and a runbook that says "launch the stack".
Those two numbers are set by two different edges. Nightly copies lose up to a day, and provisioning an environment under pressure takes hours.
Choose the topology that clears both numbers.
Components — tap one, then tap a slot on the diagram
?Fifteen minutes to recover and one minute of data loss are contractual. Those two numbers are set by two different edges.
Boundaries, outermost first: Region A · live: App tier, Primary Region B · standby: an empty slot for the compute in region B, an empty slot for the data in region B Outside every boundary: DNS (health-checked) Connections: DNS calls App tier (step 1) DNS calls compute in region B — on failover (step 2) Primary publishes to data in region B — continuous (step 3)
DNShealth-checked
App tier
Primary