Scaling MySQL at Wordpress.com
summary
wordpress.com runs about 50 million sites backed by roughly 500 million tables across 120 shards and ~2,200 servers (≈500 DB servers), serving hundreds of thousands of queries per second. Operational fixes include switching MyISAM/InnoDB, sharding by site id, a HyperDB open source dropin for routing and failover, SSDs, LVM snapshots plus mysqldump for restores. Automated rebalancing moves ~400k sites/day, backups cost ~10% of infra and restores average 22 minutes against a 1‑hour SLA.
AI-written summary · about
chapters (7)
- 0:00Start
- 1:51Tables
- 4:35InnoDB Data Dictionary
- 11:08Static Hashing
- 12:37Dynamic Assignment
- 15:28Performance
- 16:26Backups
related talks
26mOperationalizing 19 years of infrastructure via MySQL-as-a-Service
53mMySQL DevOps
29mScaling Rails to 2M MySQL requests per second
40mMySQL at Scale: Booking MariaDB to MySQL Migration
26mHow MySQL Configuration Tuning Boosts App Performance
50mMySQL 8.0 at 4. Where are we now?
35mState of the Dolphin
25mData in the Cloud