Archives of the TeradataForum
Message Posted: Tue, 14 Aug 2001 @ 11:28:01 GMT
My current understanding is that Teradata is designed to do two-table-joins and nothing else.
I am not sure that it would be a good idea to force the optimiser to resolve the N-1 joins in a single step. One or many of the PI could be non unique and create n-m relations. The numbers of rows could be a different for every table. Some Where-conditions might be selective and other not. By starting with the right table could be the best thing to reduce the overall impact. If you like to pass it in one step you have to do the same evaluation as in the N-1 approach but it might be much more difficult to decide which is the best starting point.
My understanding of your requirement would be that every table will only scanned once but therefore it is more likely that every table has to be scanned complete. This not necessarily the case in the N-1 join solution (if you are using some selective Where-conditions and some small tables).
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|