Archives of the TeradataForum
Message Posted: Fri, 22 Nov 2002 @ 18:05:45 GMT
<-- Anonymously Posted: Friday, November 22, 2002 12:50 -->
Does anyone know of a definitive answer for the following statistics collection issues:
1) Suppose statistics are currently being collected at the index level for a non unique secondary index. The index is a composite index containing two columns. Occasionally queries specify only one of the columns in the where clause instead of both. What would the best and least redundant option be that would help the occasional queries as well as the more frequent queries that specify both column values :
- collect statistics only at the individual column level for each column in the composite index
- collect statistics only at the index level
- collect statistics at both the index level and at the column level for the specific columns involved
2) For single column indexes, some statistics have been previously captured at the column level instead of the index level. We intend to start collecting these at the index level. An obvious answer would be to drop the column stats before collecting the index stats - but what would the implications be if we didn't and the column level stats went stale but the index level stats were kept up to date?
Any insights are appreciated!
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|