Introduction to TokuDB storage engine for large data sets
summary
Percona reports TokuDB as a production-ready MySQL/MariaDB engine offering fractal tree compression that gives higher compression ratios with low CPU overhead and fewer SSD writes, and an auto-sized cache that defaults to half system memory so fewer knobs need tuning. Benchmarks show faster bulk loads (the 1 billion rows insert benchmark completes far quicker with far fewer writes) and Linkbench workloads that don’t degrade over time. Conversions use PT-OSC or switch-over on a slave, and online schema changes tens of times faster than InnoDB.
AI-written summary · about
chapters (21)
- 0:00TokuDB introduction
- 2:13Why TokuDB stands out
- 3:39Minimal config and tuning
- 4:22Compression in TokuDB
- 6:34Storage for high data velocity
- 8:29TokuDB with SSD/flash
- 10:18SSD cost and endurance
- 11:47Mixed workloads handling
- 13:14Fractal trees stability
- 13:56TokuDB vs InnoDB load
- 15:21Compression impact on load
- 16:45LinkBench benchmark
- 18:16Read block size optimization
- 18:57Update operations advantage
- 20:24Scan performance with compression
- 21:52Compression strategies
- 22:35Choosing TokuDB over InnoDB
- 23:37Schema changes efficiency
- 24:44Large data sets definition
- 25:48Migrating to TokuDB
- 27:37Q&A
related talks
42mTokuDB internals
01h 57mPercona Community Live Conference - Day 2 - Open Source Databases Conference
55mThe Future of MySQL
47m10 Reasons Why the TiDB Database Should Power Your Next Development Project
29mDistributed SQL How TiDB Modernized Data Management
29mDistributed SQL: How TiDB Modernized Data Management
54mBreakout Session
26mLeveraging a Distributed SQL Database at Tokopedia: Best Practices & Lessons Learned