DB level granularity in replication

[Aleksander Kujbida]

Hello,
I need to keep a copy of each of several remote databases in my office. The office copy and remote copy are both masters (no problems there). The remote dbs each connect once daily by phone line to replicate two ways. The office master has only one binlog, thus writes to all dbs at the office will be sent over the wire to each of the remote dbs, even though most of these writes are not applicable to every remote db. It's only thousands of kb per db, but many remote dbs, so I need a way to reduce the phone line traffic. My first thought is multiple binlogs: one for each db. Is there such a possibility or work-around, or some other approach for creating a 'master db' rather than a 'master server'?

The MySQL approach to replication seems to restrict the idea of master to the whole server rather than a database. This forces admins to run a separate server for each db when db-level granularity is required in replication. I can't run 43 servers ;-) ! I'd love to contribute, but don't have the expertise to do so in a timely manner.
Sincerely,
Aleksander