RTO and RPO
also called Recovery Time Objective, Recovery Point Objective
How long recovery may take (RTO) and how much data may be lost (RPO), the two numbers that determine the cost of a resilience design.
RPO is measured backwards from the incident: an RPO of one hour means losing up to an hour of committed data is acceptable, which permits hourly backups. An RPO near zero requires synchronous replication, which costs write latency on every transaction forever.
RTO is measured forwards: an RTO of four hours permits restoring from backup; an RTO of minutes requires a standby that is already running.
Both are business decisions with engineering price tags, and the useful conversation is the one where the price tags are visible. Stakeholders asked "how much data can you lose?" always answer none; asked "would you accept a 15% latency increase on every write, plus a second region, to go from five minutes to zero?" they answer differently and more usefully.