Billions of Records in MySQL: Lessons Learned
summary
Claims a reproducible full-text boolean bug that makes some MATCH ... AGAINST queries take a different planner path and time out, demonstrates a production data breach search engine that indexes billions of rows and works, and shows concrete fixes: tune InnoDB buffers, use partitioning, avoid row-by-row INSERT in favor of select into outfile / load data infile for ingest and backups, dedupe with external sort -u to avoid costly ALTER TABLE copies, and kill runaway queries via SHOW PROCESSLIST and KILL.
AI-written summary · about
chapters (13)
- 0:00intro
- 0:14understanding full text search queries
- 1:28identifying and fixing query bugs
- 2:08speaker introduction and background
- 3:35agenda and performance basics
- 4:32optimizing mysql configuration
- 6:31choosing and optimizing storage engines
- 8:11schema optimization principles
- 9:59handling large datasets efficiently
- 15:07advanced features and techniques
- 15:57backup strategies for big data
- 21:25tools and best practices
- 24:10conclusion and feedback
related talks
34mMySQL and InnoDB Performance for the Rest of Us
33m7 or so deadly MySQL performance sins
47mMySQL Indexing Best Practices
42mThings every developer absolutely, positively needs to know about database indexing
21mWriting High Performance SQL Statements
44mConfessions of a DBA: worst and best things I've done in production
59mBoosting MySQL Performance
01h 23mBuilding Data Breach Search Engines with MariaDB