The only circumstance in which I've used first_rows is in ETL processing, where the ETL has been on a different machine that has proved to be the performance bottleneck. In that case I wanted the ETL server to start processing as soon as possible, and used FIRST_ROWS. Other than that, it's ALL_ROWS every time.Quote:
Originally posted by mike9
[B]Who of you ever work on a larger DWH project? Because in a DWH envrionment I realy don't see why the optimizer mode "all_rows" should be a problem.