Archives of the TeradataForum
Message Posted: Sat, 17 Jun 2000 @ 13:15:28 GMT
At the risk of stirring a hornets nest, I'd like to ask if anyone else has noticed some serious regressions in the optimizer with V2R3?
I can't recall, but it seems to me that we used to be able to do a quick check for the existence/validity of a column (especially in a view) by doing a "select c1 from v1 where 1 = 0" and the optimizer didn't go and scan the view. I am working with a tool that issues these kinds of queries and V2R3 is scanning the view! It is a pretty complex view, but...
I have seen other problems with small table - large table joins. I am seeing redistributions to the key of small reference tables where 90% of the rows in the large table have one value in the joined column - resulting in particularly skewed queries. Yes, we re-collected stats.
With all the new SQL features and join types (and resulting optimizer complexity) maybe it is time to give us some more tuning tools? Like static plans, ability to tell the optimizer to spend more time looking for a good plan, the ability to safely adjust statistics, and maybe even hints ...
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|