Archives of the TeradataForum
Message Posted: Thu, 05 May 2005 @ 09:55:51 GMT
<-- Anonymously Posted: Wednesday, May 04, 2005 20:03 -->
We had a problem with inserts and particularly updates to a table being very slow. We tracked it down to the fact it was defined as a SET table but did not have a unique index - so there was an overhead in enforcing the uniqueness every time an insert or update was done. It was easily fixed once identified.
I wrote a query to list all the SET tables in the system without unique indexes - to see if we might potentially have the same problem elsewhere - most of the tables picked up were too small to be of concern; but one which was not of insignificant size was DBC.AccLogTbl. It has a non-unique primary index of LogonDate, LogonTime.
Should this be of concern? We have logging turned on for all users on nearly all databases for all statements with SQL - and the AccLogTbl was gotten up to about 10Gb.
Do you think the uniqueness checking on this table would be putting any noticeable delay into queries?
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|