Infrastructure for a distributed database?

I'm building a system at the moment which has local SAP DB instances at
various offices around Australia, connecting to a Headoffice database. 
Links are over the public internet.  The branch offices need to
replicate data back to head office, while head office needs to replicate
other data out to some or all of the branch offices.

My current design is along the lines of;
triggers on the tables to replicate, to kick off the SAP loader
A python script to package the resulting data and pass it to a jabber
server using XML-RPC over jabber.  
Jabber server to server comms to give a reasonably robust queueing
mechanism
A python script called at the other end by the RPC, which unpacks the
data and calls the loader to upload it into the receiving database.

This is coming together quite nicely, I have a proof of concept
replication working well.

But it just occurred to me that someone on the list may have already
done something like this.  So I thought I'd ask for ideas.

Has anyone solved the problem of selective database replication over an
insecure WAN any other ways?

Can anyone see any gaping holes in my solution?

Any critiques, ideas, general comments gratefully received.

Thanks

Alan Graham