Archives of the TeradataForum
Message Posted: Wed, 08 May 2002 @ 13:07:36 GMT
Here's my scenario:
Day 1 we run a query that uses a value ordered index on a date column and query runs fine.
Day 2 we run our weekly load process that adds approx 3.5 million rows to a 500 million row table. Rerun query for the new weeks worth of data, but optimizer doesn't use value ordered index this time, but chooses a total different, less efficient path and query dies on spoolspace error. I recollect stats for the table and now optimizer is back to using the value ordered index again and query runs fine.
QUESTION: I should not have to re-collect stats when less than 1% of the table is changed or is this different for value ordered indexes?
It appears that the optimizer doesn't know about the new weeks worth of data in that value order index.
Any and all suggestion welcome.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|