Blue-green deployment eliminates deployment downtime by running two identical production environments and switching traffic between them instantly — a genuinely elegant solution to the “deploy without an outage” problem.
1. "Blue" environment is currently live, serving 100% of production traffic 2. Deploy the new version to "Green" (currently idle) - test it thoroughly 3. Switch the load balancer/router to point at Green instead of Blue 4. Green is now live; Blue sits idle, ready as an instant rollback target 5. If Green has problems, switch back to Blue immediately - true instant rollback
Why This Beats a Rolling Update
A rolling update replaces servers gradually, meaning old and new versions run simultaneously during the rollout — fine for most changes, but genuinely risky for anything with a backward-incompatible database schema change or API contract change. Blue-green’s instant, all-at-once cutover avoids that mixed-version window entirely.
The Real Cost
Blue-green requires running two full production environments simultaneously during the transition — double the infrastructure cost, at least temporarily. It also doesn’t solve database migrations cleanly on its own; a schema change still needs to be backward-compatible with both the old and new application versions during the cutover window, or the rollback safety net breaks.