AI-Assisted Post-Mortem Analysis for PostgreSQL
summary
AI-assisted post-mortem claims to turn bulky Postgres logs into actionable RCA and tuning steps using monitoring + LLMs; a live SQLdo.ai demo connects to an AWS RDS Postgres, creates an advisory lock blocking, and returns specific, runnable SQL plus fixes like remove advisory_lock calls, replace SELECT , and add LIMIT. Stack mentioned: PGAdmin, Grafana, Kafka, Bedrock/LLMs and Datadog-style monitors. Case study reports a 50% execution time reduction* after applying AI-driven optimizations.
AI-written summary · about
chapters (20)
- 0:00Introduction and speaker intro
- 1:30Agenda and talk overview
- 4:42What is post-mortem analysis
- 6:29Why AI in database performance
- 7:34AI-assisted workflow explained
- 9:03PostgreSQL performance bottlenecks
- 10:49AI architecture and capabilities
- 11:30Data collection for AI insights
- 12:16Anomaly detection with AI
- 14:02Predictive optimization tools
- 16:12Dependencies for AI implementation
- 16:32Case study success story
- 20:05Live demo introduction
- 21:15Demo setup and blocking creation
- 22:16SQL AI tool walkthrough
- 24:44Query optimization suggestions
- 26:09AI as a performance tool
- 27:02Demo recap and steps
- 27:42Key takeaways and call to action
- 29:29Q&A session
related talks
28mAI-Driven Incident Response: From Alert to Postmortem
34mAnalytics and AI with PostgreSQL - Mission Possible!
34mAnalytics and AI with PostgreSQL - Mission Possible!
50mLeveraging AI as a PostgreSQL DBA
35mPostgres in Artificial Intelligence
17mHow We Debug 1000s of Databases with AI: Lessons from an AI-Assisted Database...
27mBuilding safety tooling for risk-free AI tuning of Postgres
07mAccelerating RCA with AI