Production Postgres falls over for one of three reasons: a missing index, the same query running ten thousand times when it should run ten, or a join that decided today is the day to seq-scan a 50-million-row table. The trick is knowing which one. Without an APM like DataDog, most