A new column sounds simple. It is not. In relational databases, adding one at scale can create downtime, lock tables, or trigger costly index rebuilds. The impact depends on schema change strategy, storage engine, and the live workload.
In PostgreSQL, an ALTER TABLE ADD COLUMN command is fast if the column has no default value. With a default, older versions rewrite the entire table, which can cripple performance. In MySQL, ALTER TABLE often runs as a blocking operation unless you use ALGORITHM=INPLACE or tooling like gh-ost. For distributed databases, a new column change may require coordinated rolling updates and schema syncing across shards.
Design choices matter. Use nullable columns when possible to avoid backfilling under load. For non-null defaults, insert values asynchronously in batches instead of during the schema change. Always test on a full-scale dataset mirror to expose execution time and lock contention.