Postgres Summit US2026

Most Postgres users know EXPLAIN ANALYZE exists. Far fewer know how to read it confidently under pressure when a query is melting your production database at 2 AM.This talk cuts through the noise and builds a practical mental model for interpreting query plans: Reading the tree: how Postgres executes plans inside-out and right-to-left, and why that matters Cost vs. actual time: when the planner's estimates go wrong and how to spot the divergence Node-by-node breakdown: Seq Scan vs. Index Scan vs. Bitmap Heap Scan when each appears and what it tells you The usual suspects: bad row estimates, missing indexes, DISTINCT misuse, and over-eager nested loops Tools that help: EXPLAIN (ANALYZE, BUFFERS, FORMAT JSON), auto_explain, and pg_stat_statements as a triage layer Every concept is illustrated with real queries drawn from production workloads not toy examples. Attendees will leave with a repeatable debugging checklist they can apply the same day.