database-migrations

Zero-downtime database migrations allow you to add columns without table locks by separating schema changes from application deployment. This involves adding nullable columns, backfilling data asynchronously, and ensuring your ORM like Prisma or Django handles the transition smoothly before enforcing constraints.

A zero-downtime database migration is a deployment strategy that evolves your database schema without interrupting service. It involves breaking large schema changes into smaller, backward-compatible steps that prevent table locks and data loss, ensuring continuous availability during production deployments across PostgreSQL and MySQL.

Yes, this guidance covers tooling support for popular ORMs and migration tools including Prisma, Drizzle, Django, and golang-migrate. It provides specific patterns, anti-patterns, and safety checklists for managing schema changes and rollbacks across these frameworks.

To safely rollback a data migration in production, ensure all changes can be safely reversed by planning a clear rollback path before deployment. This involves creating backward-compatible schema changes and writing specific down-migrations that restore previous states without data loss.

Common anti-patterns when changing a database schema include adding columns with default values that lock large tables, running data backfills synchronously during deployment, and making non-backward-compatible changes that prevent safe rollbacks in production environments.