Blue-green deployment
A deployment strategy with two identical production environments — "blue" (current) and "green" (the new version). You deploy and test in green, then switch all traffic there at once. If something goes wrong you switch back instantly. Zero-downtime with instant rollback.
Mechanic: blue runs live; you deploy the new version in green (a parallel, identical environment), test it at leisure, and then repoint the load balancer/router from blue to green in one snap. Blue is left running for a while as an instant fallback — if a problem arises you switch back to blue immediately (no redeploy, just a routing change). Win: zero downtime on switch, and the fastest possible rollback. Trade-off: expensive (two full production environments at once), and database schema changes are tricky (both versions must work against the same database during the transition). Distinct from canary (a gradual rollout to a fraction of users) — blue-green is an abrupt all-or-nothing switch. A classic deploy strategy. Related to canary deployment and rolling update.