Archives of the TeradataForum
Message Posted: Wed, 28 Aug 2002 @ 09:11:22 GMT
Typically a complex query would go through many execution steps.
For example, if we had a query that was made up of 20 steps, joining say 15 tables together. If you are selecting 20 fields from these tables, 10 of which are compressed. It is very possible that early on in the query execution process, many of these fields will be brought into a Spool file, and then passed from one Spool file to another. If only 2 of these compressed fields are required for joining, sorts etc, then you have 8 fields which are being unnecessarily decompressed.
There are two big benefits from compressing fields.
1. Saving Disk Space
2. Performance of Table Scans
Both of which are lost as soon as a table is retrieved as part of the query process.
I can not see why Teradata can not perform the full query execution with the fields compressed. After all, the compression used is only bitmap encoding, which is what a computer does best.
Certainly this would be something worth looking into for future releases of Teradata.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|