DB Migration Strategy
Overview
Use this skill to evolve schemas in production without service disruption or hidden data corruption.
Scope Boundaries
- Schema changes affect live read/write paths.
- Multiple service versions must coexist during rollout.
- Data backfill or contract transition is required.
Core Judgments
- Migration pattern: expand-contract, dual-write, shadow-read, or phased cutover.
- Compatibility window duration and supported versions.
- Backfill approach and execution safety.
- Rollback semantics and data reconciliation strategy.
Practitioner Heuristics
- Prefer additive/compatible changes before destructive cleanup.
- Separate schema deployment from application behavior switch.
- Dual-write without reconciliation plan is a corruption risk.
- Large backfills need throttling and progress observability tied to business impact.
Workflow
- Define change classes: additive, transitional, destructive.
- Sequence schema and application releases for compatibility.
- Plan data migration/backfill and failure handling.
- Define cutover trigger and rollback decision points.
- Execute deprecation/removal only after compatibility window closes.
- Document residual migration debt and retirement deadlines.
Common Failure Modes
- Breaking changes shipped before all consumers are updated.
- Backfill jobs compete with production traffic and cause incidents.
- Rollback plan restores code but not data semantics.
Failure Conditions
- Stop when compatibility window cannot be supported operationally.
- Stop when rollback semantics for migrated data are undefined.
- Escalate when migration risk exceeds release risk tolerance.