Canary Release Testing
Canary release testing validates a new version by rolling it out to a small percentage of real production traffic first, closely monitoring for errors or performance regressions, before gradually increasing the rollout to the full user base — catching problems while the blast radius is still small.
The name references the historical practice of canaries in coal mines serving as an early warning system — a small subset of real users (often 1-5%) unknowingly serves the same role, absorbing the risk of a bad release before it reaches everyone, with automated monitoring watching error rates and performance closely enough to catch problems within minutes.
The testing discipline here is largely about the monitoring and rollback automation, not manual test execution — defining clear, automatic rollback triggers (error rate exceeds X%, latency exceeds Y) ahead of time is what makes a canary release actually safe, rather than just hoping someone notices a problem manually in time.
