Navigating the Challenges of CI /CD for Relational Database Schema Changes
summary
Relational schema changes cause long locks, data loss and downtime when DDL runs in production; simple diffs or desired‑state tools (Liquibase, Flyway, EF migrations, or great) can hide that risk. The recommended approach is classification (compatible vs breaking, online vs offline), run changes against an ephemeral pre‑prod snapshot with a pre-prod ephemeral snapshot quality gate that fails on schema modify locks, and convert offline breaking edits to online procedures by create new table and batch-migrate with triggers, swap tables, and delay dropping old columns.
AI-written summary · about
chapters (10)
- 0:00Introduction and background
- 3:14Database challenges in DevOps
- 10:09Historical CI/CD for databases
- 17:14Tools and automation limitations
- 25:56Database migrations vs desired state
- 31:40Deployment risks and downtime
- 36:16Database change matrix explained
- 42:42SQL internals and locking
- 53:20Online vs offline changes
- 57:30Best practices and summary
related talks
33mDatabase Schema Migrations with Zero Downtime
22mLessons DevOps Can Teach Us About Managing Internet Scale Schemas
31mHelp! My Database Is The Bottleneck
27mStrategies for Reliable Database Schema Migrations
31mDevOps Practices for the Database Team
11mLightning Talk: Continuous Delivery…except for Database Changes
45mEnabling Continuous Delivery with Database Practices
37mDatabase Schema Migrations with Zero Downtime