kimtth Goal: schema changes that are reversible, ordered, and safe in production.
Use for:
- adding, altering, or dropping columns and tables
- coordinating schema and code deploys
- avoiding locks and data loss on large tables
Workflow:
- Make each migration small, ordered, and version-controlled.
- Prefer additive changes; split risky changes into steps.
- Use expand-then-contract for renames and type changes.
- Backfill data in batches to avoid long locks.
- Deploy code that tolerates both old and new schema.
- Verify on a production-like dataset before rollout.
Expand/contract:
- add new column/table, deploy dual-write code
- backfill, then switch reads
- remove old column only after nothing uses it
Rules:
- never edit an already-applied migration; add a new one
- avoid long-held locks on large tables
- make migrations reversible or document the rollback
- test against realistic data volume, not an empty schema