role-architect:migration-planner

Migration Planner

When to use

• Planning an incremental migration from a legacy system using strangler fig
• Choosing between shadow mode, dark launch, and graduated cutover strategies
• Designing ETL, CDC, or dual-write data migration approaches
• Tracking feature parity between legacy and new systems during a migration
• Defining rollback procedures and automatic rollback triggers per migration phase
• Applying zero-downtime techniques (blue-green, rolling, expand-and-contract)

Core principles

1. Facade before cutover — no traffic switches without a routing layer in place and tested
2. CDC over dual-write for anything longer than a sprint — dual-write consistency problems compound fast
3. Rollback must be tested before go-live — an untested rollback plan is not a plan
4. Each phase independently deployable — never combine two risky changes in a single phase
5. Expand-and-contract for schema changes — the only reliable zero-downtime database migration pattern

Reference Files

• references/​migration-strategies-and-execution.md — strangler fig pattern with facade design, parallel run modes (shadow/​dark launch/​graduated cutover), ETL and CDC data migration techniques, dual-write trade-offs, feature parity matrix and API-level tracking, phased timeline (Phase 0-4), and zero-downtime techniques (blue-green, rolling deployment, expand-and-contract schema migration, API versioning, feature flags)
• references/​rollback-and-risk-mitigation.md — rollback procedure requirements (data rollback, traffic rollback), rollback trigger thresholds (error rate, latency, data discrepancy), risk identification per phase with likelihood/​impact assessment, common mitigations (feature flags, canary deployments, automated triggers), and stakeholder/​team/​customer communication plan with post-migration retrospective