Archives of the TeradataForum
Message Posted: Thu, 06 Feb 2014 @ 22:15:29 GMT
Subj: | | Re: NULL values in the ObjectColumnName field of DBC's DBQLObjH |
|
From: | | Frydryszak, Marek |
ObjectColumnName is always filled in for PI - in case of multicolumn PI we get few similar rows one for each PI column.
For multicolumn secondary indices we get one row with null in ObjectColumnName. But not always :) e.g. Primary key (actually secondary index)
is treated similar to PI.
For single column secondary indices we sometimes get column name sometimes null... Based on my observations null means, that index was used
in tricky way. E.g. query like "sel ID,count(*) from A group by 1" may be faster executed scanning SI based on column ID (if exists) instead of
scanning full possible wide table A.
|