The thing that is impractical to do on premise is "transparent scaling" say your load increases and you need a way bigger box that's the kind of problem the ser

The thing that is impractical to do on premise is "transparent scaling" say your load increases and you need a way bigger box that's the kind of problem the service can solve (take slave offline remount storage on a bigger box catchup on replication switch to slave repeat for former master). We are at the early stage just investigating if we should move forward so @andre1sk on twitter would prob the only place :)