Archives of the TeradataForum
Message Posted: Mon, 04 Nov 2002 @ 18:58:22 GMT
I suspect that, as a few people have said, you have very skewed NUSI. Often a lot of Nulls or Zeros for default or missing values.
The bad news is that this is the way the data looks.
Sometimes as a work around we substitute a well distributed column's value as a negative number in place of the Null or Zero. This allows you to easily screen out for these negatives.
The good news is that in V2 R5 which will be out in December, you will be able to build sparse NUSI. Which will allow you to not build the index rows for these very lumpy values and which should improve your performance substantially.
This option was specifically designed to address the issues you are facing.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|