What is a rollback plan?
A rollback plan, also called a back-out plan, is a standard part of a change request. It is written and approved before the change starts and answers three questions: what counts as failure, who decides to roll back, and exactly how the system gets back to where it was. Without one, a failed change tends to become hours of "fix-forward" troubleshooting in production while users wait.
What a rollback plan contains
- Triggers: measurable failure conditions, such as users unable to sign in, an application that will not launch, stop errors after a driver update or a failed validation check, plus a decision deadline inside the maintenance window.
- Authority: who approves a rollback and who carries it out, and confirmation that person has the rights needed.
- Prerequisites: a verified backup, the previous installer or driver version, an exported copy of the configuration, and a snapshot or restore point taken before the change.
- Steps: the exact actions in order, including dependencies. If an application upgrade also updated a runtime or plug-in, the plan says whether to revert those too.
- Verification: the same success checks used to approve the change, which must pass again after the rollback.
- Communication: who tells the service desk, users and stakeholders that the change was reversed.
- Documentation: what was rolled back, when and by whom, recorded in the change record and related tickets.
Rollback plan, backup plan and sandbox testing
These three work together. Sandbox testing tries the change first in an isolated environment, such as a spare PC, a test virtual machine, a lab network or a test cloud tenant. That is also the place to rehearse the rollback and prove it works. The backup plan protects data and settings so there is something to restore. The rollback plan is the procedure that uses those resources to undo the change. A rollback that depends on a backup nobody verified is not much of a plan.
Common rollback methods on Windows workstations
| Change | Typical rollback |
|---|---|
| Driver update | Roll Back Driver in Device Manager, or reinstall the previous version |
| Quality update | Uninstall it from Windows Update history, or from the Windows Recovery Environment if Windows will not start |
| Feature update | Go back to the previous version, which Windows allows only for a limited time after the upgrade |
| Settings change or software install | System Restore to a restore point taken before the change |
| Application upgrade | Reinstall the previous version from a saved installer and restore its configuration files |
| Major or widespread change | Restore from backup, revert a VM snapshot or reimage |
Rollback plan vs rollback
A rollback is the act of reverting, and in databases the word also means undoing an uncommitted transaction automatically. A rollback plan is the change management document that makes the reversal fast and predictable. It exists before anything breaks.
Rollback plan on the CompTIA A+ exam
Core 2 (220-1202) objective 4.2 covers change management, where the rollback plan sits alongside the backup plan, sandbox testing and naming the staff responsible for each step. If a question asks what to do when a change causes problems, think "follow the rollback plan," not "keep troubleshooting in production." If it asks what must be documented before a change is approved, the rollback plan belongs with purpose, scope, risk analysis and affected systems. If it asks where to test both the change and the rollback, think sandbox.