Building an Analytic Extension to MySQL with ClickHouse® and Open Source
summary
Argument: move analytics out of MySQL because a 170 million row aggregation runs 170M rows: MySQL 9 min vs ClickHouse® 0.5s and another average-by-year takes MySQL 4 min vs ClickHouse® 0.674s. Implementation: use MySQL Shell and mydumper -> ClickHouse loader for a fast initial dump, then CDC with Debezium/Kafka Connect -> ClickHouse sink and checksum validators. Result: initial bulk load and queries are excellent, CDC works in production but needs fixes for replica failover, schema-change coordination, and out‑of‑order Kafka partitions.
AI-written summary · about
chapters (13)
- 0:00Start
- 6:01Query comparison (MySQL/ClickHouse)
- 10:01Signs that MySQL needs Analytic Help
- 16:07Replication Setup
- 19:031. Initial Dump/Load
- 21:51Validate Data
- 23:55Setup CDC Replication
- 25:14Final step - Deploy
- 26:23Final step. Monitor
- 27:51Challenges
- 28:27MySQL Master Replication
- 32:54Replicating Schema Changes
- 39:06Project roadmap and next Steps
related talks
30mReplicating Data from MySQL to ClickHouse
55mBuilding Fast ClickHouse Analytics for MySQL Transaction Data Using Kubernetes
40mData warehouse or conventional database: Which is right for you?
30mSQL Row Store vs Data Warehouse: Which Is Right for Your Application?
42mClickHouse: what is behind the fastest columnar database
04mRoad to Percona Live 2022 - Robert Hodges
30mBuilding For Fast
51mTheory and practice of using ClickHouse in real applications