Archives of the TeradataForum
Message Posted: Fri, 30 Jan 2015 @ 22:24:35 GMT
<-- Anonymously Posted: Friday, January 30, 2015 17:08 -->
With the variety of statistics options now available any legacy code and tables should be reviewed on a case-by case basis.
For example, before you go down the rolled up single scan statistics path I recommend you also investigate use of the simple 'COLLECT SUMMARY STATISTICS ...' route as well.
Our experience with the former is that of spool and containment issues on large tables whereas simply collecting the 'instant' SUMMARY statistics has saved us millions of CPUs with no loss in throughput. In many cases this has speeded up processes dramatically. Explain query paths are much the same.
This is especially true on intermediate work tables.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 24 Jul 2020|