Part 2 of a series revisiting Evolutionary Database Design twenty years later, focusing on how Databricks Lakebase's copy-on-write branching removes long-standing constraints on database development practices. The post details 11 practices (7 original from 2003, updated, plus 4 new ones) enabled by instant, zero-copy database branching. Key workflows include per-PR branch creation via GitHub Actions, automated schema diffs posted to PRs, branch cleanup on merge, idempotent migration scripts, destructive testing on ephemeral branches, and A/B schema exploration on parallel branches. Concrete GitHub Actions YAML examples and CLI commands are provided throughout, showing how per-developer and per-PR database isolation becomes routine rather than aspirational.
