MySQL Backups
summary
Booking.com describes a production MySQL backup system that uses a dynamic config generator feeding a Bacula director, allocates storage by free space and previous backup size (submit largest first), takes LVM/ZFS snapshots of dedicated replicas, protects copies with NetApp SnapLock immutability, stores binary blobs with erasure coding (MinIO mentioned), and runs continuous restore tests by spinning recovered MySQL instances in Kubernetes restore testing to validate restores and reporting. Operational goals: automation, retry/prioritization, and observable audit tickets.
AI-written summary · about
chapters (14)
- 0:00Start
- 0:01Mysql Backups at Booking Com
- 2:14What Do People Want from Their Backups
- 4:28Compliance and Best Practices
- 5:20Immutable Backups
- 20:00Non-Temporal Backup Scheduling
- 21:32Current Setup
- 22:45The Backup Sources That We Use
- 24:10Config Generator
- 25:21Storage Daemons
- 35:41How Do We Find Storage
- 37:57Restore Testing
- 41:11Error Handling
- 48:14Backup Times
related talks
46mLock, Stock and Backup: Data Guaranteed
50mWeb-Scale Backups for MySQL
23mRunning MySQL on K8s. Deployment, Scaling, Backups and advanced configuration
43mBacking up Nextcloud-Consistent Backup and Restore of Application Data
57mThe Evolution of a MySQL Database System
36mCreate your Own MySQL-as-a-Service that Runs Anywhere Using Kubernetes Operators
34mSailing the Database Seas: Applying SRE Principles at Scale
26mOperationalizing 19 years of infrastructure via MySQL-as-a-Service