Archives of the TeradataForum
Message Posted: Thu, 03 Oct 2002 @ 13:51:35 GMT
As I understand the problem, Claybourne's use of TWO views on the BaseDim is forcing TWO passes through the BaseDim table, one for Elephant and one for purple. Then Elephant Spool is joined to the BaseFact table (without Purple) on a single column, this could have very poor performance. Then the results of Elephant|BaseFact are joined to Purple. Again, this can be a very poor performer.
I would except the Elephant|Purple combination should produce one row. We need to have this step performed prior to the BaseFact join. I would also expect joining the base tables, BaseFact and BaseDim , with the two filters on BaseDim would produce the preferred results. (one pass through BaseDim with both filters, producing one row, then BaseFact, very fast).
I have my thoughts one how best to solve this problem.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|