|
Archives of the TeradataForumMessage Posted: Wed, 25 Apr 2012 @ 09:20:11 GMT
Hi, Typically, you'd only collect stats on the individual columns if you have queries that only select/join using one of the columns AND not having such statistics makes the plan a 'bad one'. I would always have stats on the combination as this is the PI. I doubt if there's much difference between how you handle this for a big or small table except that remember that for a (very) small table the part of the plan that processes that table will often not change. Think of a small reference table, often with a PI on a code column. Typically, a plan involving this table will include a duplication of all/part of the table and then a product join with another table/spool file. Single/multi-column stats typically won't change that plan. Cheers, Dave Ward Analytics Ltd - Information in motion (www.ward-analytics.com)
| ||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||
Copyright 2016 - All Rights Reserved | ||||||||||||||||||||||||||||||||||||||||||||||||
Last Modified: 15 Jun 2023 | ||||||||||||||||||||||||||||||||||||||||||||||||