No, I haven't confirmed that - the documentation states that queries with remote references as well as local become distributed queries, and that parts of distributed queries are separated into chunks and sent off for remote execution where possible. My main reason for asking was to find out how good the optimiser was in practice - I could run a couple of simple test cases, but I need to know whether it'll be able to separate remote statements out of more complex queries. If it all suddenly blows up and executes sequentially, I probably can't take that performance hit. Asking if there are any common pitfalls (or dos and don'ts) for the optimizer seems prudent given this.

Views are a good idea for explicitly telling Oracle to execute a certain part of a query remotely. One misgiving I have is that the queries aren't likely to be the same each time, and are created dynamically - how expensive (and indeed possible) would it be to create a view on each remote server from the 'master' server and then execute the SQL, each time? Is there a better way to do this?

As you may have guessed I haven't had a great deal of experience with Oracle, so I'm just trying to get a feel for the standard ways of handling distributed remote queries.

Thanks for the help so far!