will be broken up and routed to the appropriate machines, executed in parallel, the resultset will be recombined, and streamed back to the client. The user never has to worry about sending queries to specific nodes— everything happens automatically behind the scenes.