What is Rollback?

What is a Rollback?

A rollback returns a system to a previous known-good state when a new release, configuration, or data change causes unexpected problems. The goal is to limit the impact of the faulty change and restore service quickly.

In a web application, rollback may be as simple as redeploying the previous container image. When database schemas or irreversible data transformations are involved, the operation needs much more planning.

How is Rollback Planned?

A reliable rollback plan answers these questions before release:

• Where is the previous application version stored?
• Is the new database migration backward-compatible?
• Is disabling a feature flag enough, or must code be reverted?
• Are caches, queues, and background jobs compatible with the previous version?
• Which metric or alert will trigger the rollback decision?

In a CI/CD pipeline, every deployment should have a traceable version number, image tag, and migration history.

Business Use

Rollback is especially important for payment, order, login, and integration flows where downtime directly affects customers. Without a plan, teams may spend the incident window trying to repair the new version while user impact grows.

Sometimes the best recovery is not restoring old code, but shipping a small hotfix or disabling the problematic feature. The decision depends on data risk, user impact, and whether the change is actually reversible.