Point-in-time query tuning and observability with pg_stat_statements

on YouTube ↗25m

" summary

pg_stat_statements is cumulative by default (resets on restart or manual reset) and exposes 31 columns of stats; the method shown snapshots that view periodically into tables, computes per-statement deltas, and visualizes calls/min and exec-time in Grafana. Implementation uses pg_stat_statements output, scheduling via pg_cron or a TimescaleDB background job, storing rows in a partitioned/hypertable (default retention/compression reduces storage). Default statement buffer is ~5,000 entries; the demo pipeline works and reveals spikes and inefficient queries.

AI-written summary · about

" chapters (5)

1. 0:00Introduction
2. 1:21pg_stat_statements primer
3. 10:47How to track historical data
4. 14:42Demo of pg_stat_statements
5. 22:57pg_stat_monitor & pganalyze

" related talks

• 51mPostgreSQL Query Performance InsightsHamid Quddus Akhtar · Percona Live 2022
• 50mPostgreSQL Query Performance InsightsHamid Quddus Akhtar · PGConf NYC 2021
• 57mQuery Logging and Workload AnalysisGreg Smith · Postgres Open 2012
• 39mDeep dive into PostgreSQL internal statisticsAleksey Lesovskiy · HighLoad++ 2015
• 49mProfessional PostgreSQL monitoring made easyPavlo Golub · PostgreSQL Conference Europe 2023
• 47mIdentifying Slow Queries and Fixing Them!Stephen Frost · PGConf NYC 2021
• 50mHow To Improve PostgreSQL Monitoring and Observability With PMM - Part 2Charly Batista, Agustin Gallego · Percona Live 2022
• 42mProfessional PostgreSQL monitoring made easyPavlo Golub · pgDay Paris 2025