Capturing and Analyzing Millions...
summary
Describes LinkedIn’s query analyzer that captures and measures millions of MySQL queries with under 3% CPU overhead and no I/O or other resource cost. Compares it to MySQL slow-query log (25–35% overhead and only captures queries above a threshold) and performance_schema (10–20% overhead and complexity), and explains how their approach avoids those drawbacks to collect comprehensive query data for optimization.
AI-written summary · about
chapters (12)
- 0:00Start
- 6:45Existing approaches
- 10:42Pain Points
- 12:37Architecture
- 13:07How Agent Works
- 15:05Anonymization Examples
- 15:33How Agent computes
- 16:59Query Metadata
- 17:48How does it all work?
- 18:18Query Analyzer Overhead
- 22:38Query Load Explained
- 23:59How can you leverage this?
related talks
55mLet's Talk About MySQL Query Tuning
54mPerformance Analysis and Optimization of MySQL
21mLeveraging Continuous Production Profiling for Providing Insights into...
50mQuery Analysis in MySQL, PostgreSQL, MongoDB
43mOptimizing MySQL without SQL or touching my.cnf
32mTackling Slow Queries: A Practical Approach to Prevention and Correction
57mPerformance Analysis and Optimization of MySQL
35mBest Practices for MySQL Performance