Adding a new column sounds like the smallest schema change you can make. It isn’t. The moment you alter a table, you risk locking rows, breaking queries, or slowing down writes. In a production database, that can mean downtime. In a replicated setup, it can mean lag that piles up until a failover turns into a disaster.
A new column has to be more than an ALTER TABLE statement. You decide if it should be NULL or NOT NULL. You choose defaults carefully—remember a default value backfills rows, which can spike IO. You consider column order if your storage engine cares about it. You know that adding indexes at the same time will multiply the work the database has to do.
Zero-downtime schema changes often need a migration tool. These tools create the new column without locking the table. They copy data in batches and track progress. Some swap the old table for the new one in a single transaction. Others use triggers to keep columns in sync until the cutover.