Aginity workbench Videos

Thanks Craig. That was helpful.

Skew defines how unevenly data is distributed across the SPUs.  With any parallel processing appliance, you want data evenly distributed so that all the parallel processors are doing equal amounts of work.Broadcast means that data required by the SPU to do the work is not locally available.  This means Netezza will copy the data (broadcast it) to the SPUs that need the data.  This can be a very costly operation, and if you see a broadcast happening, especially on your fact tables, it can highlight incorrect distributions.  However, broadcasting dimensional data, which is usually much smaller, is not necessarily a problem.