# DB Migration Strategy > Schema migration strategy workflow for sequencing changes, compatibility windows, and rollback-safe rollout in live systems. Use when schema evolution impacts running services; do not use for static greenfield schemas without deployment constraints. Use when this capability is needed. - Skill: `tomevault-io/db-migration-strategy` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/db-migration-strategy` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/db-migration-strategy/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/db-migration-strategy --- # 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 1. Define change classes: additive, transitional, destructive. 2. Sequence schema and application releases for compatibility. 3. Plan data migration/backfill and failure handling. 4. Define cutover trigger and rollback decision points. 5. Execute deprecation/removal only after compatibility window closes. 6. 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. --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/kentoshimizu) — claim your Tome and manage your conversions.