I have never done this, but I think this kind of things are better handled at the DB-driver level. Indeed, Connector/J has a specific ReplicationDriver for this purpose. The Connector/NET documentation has a rather vague reference to replication:
Host, Server, Data Source, DataSource, Address, Addr, Network Address:
The name or network address of the
instance of MySQL to which to connect.
Multiple hosts can be specified
separated by &. This can be useful
where multiple MySQL servers are
configured for replication and you are
not concerned about the precise server
you are connecting to. No attempt is
made by the provider to synchronize
writes to the database so care should
be taken when using this option. In
Unix environment with Mono, this can
be a fully qualified path to MySQL
socket file name. With this
configuration, the Unix socket will be
used instead of TCP/IP socket.
Currently only a single socket name
can be given so accessing MySQL in a
replicated environment using Unix
sockets is not currently supported.