These two indicators frame every recovery plan:
- RPO (Recovery Point Objective): how much data can you lose? It is the maximum gap between the disaster and the last healthy copy. Daily backup = 24h RPO; asynchronous replication = minutes; synchronous = ≈ 0.
- RTO (Recovery Time Objective): how long can you stay down? From minutes (active standby site) to several days (full restore).
The method: set RPO and RTO per application (payroll is not the production line), then derive the architecture — each step up costs an order of magnitude more. Classic trap: a zero RPO does not protect against ransomware replicated in real time to the backup site; also keep immutable, disconnected backups. And measure the real RTO through failover tests, not on paper.