Archives of the TeradataForum
Message Posted: Wed, 30 Mar 2005 @ 17:49:59 GMT
| Subj: || || Re: Statistics Help |
| From: || || Fadel, Michelle |
Tad Warnick is here onsite and he wants to say that this diagnostic is not supported and collecting stats can be very expensive. The hierarchy
to be followed is single column high confidence, followed by single column low confidence (single column stats are not that expensive). Look at
multi column high and low, but just because stats are being evaluated, it doesn't mean that collecting them will change the plan.
In my experience, collecting multi column stats recommended by this diagnostic ran up my collect stats hours but didn't change the queries. We
ultimately dropped those multi column stats.