collecting feedback from potential customers/users has been the most important thing I've done in this past year. A few people that I talked to commented that they would like to see a backup/restore feature. This feature is just about to be released.
It's something I'm quite proud of because the backup architecture was designed to not cause any downtime to the master database. I ended up running two instances of mysqld for each server. One mysqld acting a slave/read only of the master database.
Backups are taken from the slave. So any writes to the master database will always succeed.
Restores are run directly on the master database and replicated down to the slave node. I imagine restoring will happen on initial application create time.