Disaster Recovery
also called DR
The plan and capability for restoring service after an event that takes out a whole site, region or system.
DR is defined by two numbers, and any DR conversation that has not produced them is not yet a DR conversation. RPO is how much data you can afford to lose; RTO is how long you can afford to be down. They drive everything else and they are expensive in a strongly non-linear way.
The four common postures, cheapest to costliest: backup and restore (hours to days), pilot light (core systems running small, scaled up on failover), warm standby (a scaled-down full copy running), active-active (both sites serving, failover is a routing change).
The single most important fact about a DR plan is whether it has ever been executed. An untested plan is a hypothesis, and the recurring discovery during real invocations is a dependency nobody listed — DNS, a certificate, a licence server, a runbook stored only in the failed region.