As I understand it, if you choose rolling backups then the db naming convention is different to the one for static dumps. This means switching between the two techniques requires manual intervention. With a couple of tasks to check the current database names and create a new database with the right name as precursors to running the rest, we could totally make it possible to switch deployment backup technique without actually manually creating databases. π