hot / warm / cold standby

networkingcloud

The three temperatures of a backup system: running and synchronized, running but behind, or powered off until needed.

The temperature buys recovery time: a hot standby takes over in seconds, a cold one after hours of restore. Cost scales the same way, which makes the choice a business decision wearing a technical costume.

Standby configurations describe how ready a secondary is, and the terms map to a spectrum of cost against recovery time. Hot means running and synchronized, able to take over in seconds. Warm means running but behind, needing some catch-up. Cold means the equipment exists and must be started, configured and loaded before it is useful.

The choice is an economic decision disguised as a technical one, and it should be driven by two numbers agreed with the business rather than by engineering preference: how much time may pass before service resumes, and how much data may be lost. Those two determine which configuration is required, and everything else is implementation.

The failure that recurs is a mismatch nobody noticed. A cold standby documented as meeting a four-hour objective, where nobody has measured how long the restore actually takes, is a plan that will be discovered to be wrong during its first use. And a hot standby whose synchronization has silently broken is worse than a cold one, because the readiness is assumed. Both cases are resolved by the same practice: exercise the failover on a schedule and time it.

Also known as: hot standby, warm standby, cold standby

← All glossary entries