A zero-downtime deploy checklist for small teams

Zero-downtime deployment gets treated like a big-tech luxury — something you earn once you have a 50-person platform team. It isn't. It's a checklist, and a small team can run it before every release. Schema changes go out backwards-compatible: add the new column, deploy the code that uses it, then drop the old column in a later release — never all three at once. Container updates roll out blue-green or in rolling fashion behind the load balancer you already have, whether that's an AWS ALB or Cloudflare, so old and new versions serve traffic side by side for a few minutes instead of a hard cutover. A health check has to confirm the new containers can actually reach the database before traffic gets routed to them — a container that boots but can't connect is worse than one that never boots at all. And a rollback has to be one action, triggered by an automated error-rate check, not a person watching a dashboard at midnight deciding whether tonight is the night it breaks. None of these four steps requires a platform team. They require deciding, once, that this is how releases work here — and then never skipping it because a deploy "should be fine."