I have been occupied with getting the Virtuoso cluster to fuifull functionality.  It now does insert/delete/update and select over partitions spread over multiple instances and can send co-located joins and other co-located query fragments over to partner nodes.  Of course, the more functionality is transferred per  message, the better.

Most of the code is definitely written.  Now it is a question of testing and adding optimizations here and there, also detecting distributed deadlocks.

I'll go to the SABRE conference in Leipzig this week, where we present an update on Virtuoso.  This is an updated verversion of the Virtuoso intro paper, now also at the Virtuoso web site.
 
In another month, I'll be at the RDF Relational Mapping workshop in Cambridge, MA.  There will be more updates on our results before then.