Continuous Database Administration
summary
Cloud DBs (Amazon RDS, Google Cloud SQL, Aurora) handle provisioning, replicas and backups but leave tuning and query optimization manual. An interactive psql menu and scripts plus pg_replay and cloned replicas run parallel experiments (about 10 at once on spot instances) to test index changes. One experiment cuts total runtime from 45 to 43 minutes, yields a 55% overall improvement, SELECTs ~20% faster, and much smaller indexes at the cost of slower updates after adding partial indexes.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 4:45Performance hacker role
- 9:21Current DB monitoring
- 15:47PostgreSQL bloating
- 23:01Query analysis and optimization
- 27:54Reactive vs proactive model
- 31:07Future of DBA automation
- 36:54Clouds and databases in Russia
- 38:21Q&A
related talks
43mPostgreSQL in the cloud: Theory and Practice
40mPushing past PostgreSQL’s vertical scaling limits - Running Amazon RDS
46mHow we made PG Fitter, Happier, More Productive
36mBest Practices in PostgreSQL Tuning: Navigating Key Performance Bottlenecks in the Cloud
48mDBtune: AI-driven performance tuning for all PostgreSQL flavors
46mBig databases, many servers, data center or cloud - it's all here!
32mManaging Databases in the Cloud in Broken
46mPostgres Platform "Best Practices" for the Modern DBA