Postgres Summit US2026

Most teams running Postgres in production do not have a dedicated DBA. The backend engineer owns it. Between shipping features, handling incidents, and keeping the lights on, database maintenance becomes the thing you do when something breaks.

This talk is about not waiting for something to break.

We cover the practical checklist that small teams actually need: spotting table bloat before it kills performance, reading slow query logs without a data platform, setting up autovacuum correctly for your workload, index hygiene, connection pool sizing, and backup strategies you will actually test. We also cover the monitoring signals worth paying attention to and the ones you can ignore.

The goal is not to turn engineers into DBAs. It is to give them just enough to keep a production Postgres cluster healthy with minimal overhead, and to know exactly when a situation has grown beyond what a generalist should handle alone.

Good database hygiene is not complicated. It is mostly about doing a few things consistently and catching problems early.