Yeah unless your database is not a critical component of your application, like a experiment or doesn’t have to be available feature you should almost always self operate. It pays to have control of your data.