Archives of the TeradataForum
Message Posted: Wed, 02 Oct 2002 @ 14:21:06 GMT
I have a case where (in order to make my data look like a star schema to the reporting tool) I'd like to make a single physical table (BaseDim) look like mulitple dimension tables via the use of views. Each 'View' of the BaseDim contains a unique subset of columns.
After creating these views, I ran some explains to understand how the optimizer would handle the SQL. As it turns out, each view resulted in the creation of a seperate spool file. I was hoping that the Optimizer would only create a single spool file resulting from a single scan of 'BaseDim.'
Now, in terms of recursive joins, I understand why it is important to treat each view of a table as a seperate spool file when the Optimizer resolves a query; however, in my case, the optimizer should have been smart enough to see that each of my 'views' of BaseDim looked at a unique subset of columns of the same physical table; therefore, it should have performed a single scan of the table and created a single spool file of the results.
Does anybody else agree?
|Copyright 2016 - All Rights Reserved|
|Last Modified: 23 Jun 2019|