Google have contributed some code to the Java Hibernate project (an object/relational persistence and query service): Hibernate Shards. This distributes the objects throughout multiple databases according to certain criteria: because "sometimes you simply have too much data". The documentation is an interesting read, particularly Querying and Limitations: "Distinct clauses are trouble as well. So much trouble, in fact, that at the moment we don't even support them. Sorry about that".
C. J. Date would not approve.
My brother in London has been doing a lot of work managing and programming a large grid computing project. They decided to jettison their databases and all data lives on the grid. The ORM impedance buzzword problems go away. He even claims it's faster that way. I'm rather curious how transactions work, but I suppose that's another story.
