OpenTelemetry, databases, and machine learning
summary
Captures SQL and planner output with OpenTelemetry traces and projects developer changes onto production data and indexes to catch slow queries, N+1s, and risky schema migrations before staging. Uses database execution plans plus a telemetry collector and ML/LLM analysis (ChatGPT) to produce concrete fixes and SQL/index recommendations; an example query drops from 110s to 3s after automated suggestions. Commercial-style tooling (Mattis) and rules automate alerts and prevent late-stage load tests.
AI-written summary · about
chapters (9)
- 0:00Introduction and motivation
- 5:45Database problems in modern systems
- 17:22Why tests and load tests fail
- 19:30Building database observability
- 22:45Understanding query execution plans
- 24:51OpenTelemetry for database tracing
- 27:40Early validation in development
- 29:44Automating fixes with AI
- 30:49Summary and future steps
related talks
52mGraphQL Observability with Elastic and OpenTelemetry
15mOLTP SQL Database Query Tracing and Linting
35mOpenTelemetry for Busy People
33mThe Observant Developer - OpenTelemetry from code to prod
23mOpenTelemetry Auto Instrumentation for Effortless Observability
22mEnhancing Database Observability With OpenTelemetry
38mThe State of OpenTelemetry
49mAdding OpenTelemetry to Production Apps Lessons Learned