An administrator is required to write a contingency plan

Question

Does anyone know the answer to the following question:

An administrator is required to write a contingency plan before making significant changes to a production virtual server. If the change fails, the server must be reverted to its original configuration in a relatively short amount of time. Which of the following will BEST achieve that goal?

1. Snapshots
2. Sandboxing
3. White box testing
4. Initial baseline configuration
5. Trusted OS