When you add a new column to a relational database, the database must update its metadata and, in some cases, rewrite large portions of existing data. The cost can be small or massive, depending on table size, nullability, and default values. In production systems, careless schema changes can lock tables, block writes, or trigger replication lag.
Plan the new column. Decide type, nullability, defaults, and indexing before you touch production. Understand how it will interact with ORM models, migrations, and downstream systems. For SQL databases, the ALTER TABLE ... ADD COLUMN ... statement is the standard, but its behavior varies by engine. PostgreSQL adds most new columns instantly if no default is specified. MySQL can rebuild tables. SQLite rewrites the entire file.
Test migrations in staging with production-like data. Check data integrity, query plans, and read/write performance after adding the new column. Monitor application errors, API responses, and slow query logs after deployment. Roll forward with zero-downtime patterns: shadow writes, dual reads, and background backfills if data population is required.