It is desirable to keep the configuration of DataPower objects and services under version control. The are many advantages to this approach: being able to re-install a release of an application (of which DataPower configuration could be a component), reverting to an older version of the configuration, being able to diff versions (including diffing across environments) and so on. DataPower itself has built-in support for checkpoints and configuration diffing, but it's no match for what can be done with a version control tool, especially the one that has a light-weight branching model such as Git.

You can find more details about version-controlling DataPower configuration here.