How To Improve PostgreSQL Monitoring and Observability With PMM - Part 2
summary
Percona Monitoring and Management (PMM) collects PostgreSQL metrics (Grafana front end, VictoriaMetrics for timeseries, ClickHouse for query analytics, PostgreSQL for config) with up to one-second resolution retained for months and supports push mode. Query analytics use pg_stat_statement or pg_stat_monitor and custom queries or a textfile collector add metrics per-instance. A demo shows a recursive CTE that runs in ~120 ms versus ~700 ms after query rework, dashboards for checkpoints/IO, and built-in alerts and advisors that actually fire and report useful values.
AI-written summary · about
chapters (12)
- 0:00Start
- 1:09Agenda
- 2:36The Pmm Interface
- 5:13Why Did We Change Uh Prometheus for Victoria Metrics
- 6:01Why Is Monitoring Needed
- 8:02Dashboard
- 8:30Current Analytics
- 11:09Why It's Important To Use a Standard Deviation Instead of the Average
- 13:08Live Demo
- 37:31Resolution
- 38:06Custom Queries
- 47:10Advisors and Alerts on Pmm
related talks
51mHow to improve PostgreSQL Monitoring and ...
28mThe "411" on PMM
25mpg_stat_monitor: A cool extension for better monitoring using PMM
21mEasy way of application monitoring
49mProfessional PostgreSQL monitoring made easy
30mWhy we recommend PMM to our clients
22mGetting Started with Percona Monitoring and Management
42mProfessional PostgreSQL monitoring made easy