Archives of the TeradataForum

Message Posted: Thu, 11 Apr 2002 @ 12:42:44 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Teradata always advises the use of Primary Index in the join condition and not using conditions like <>, NOT IN, Multiple OR's for Join conditions, etc.

But consider a case wherein there is a monster (big) dimension, say Customer, with many many rows and you need to give an update on rows where Customer_ShoppingAmt > 500000, Location= North_US and Status=Locked.

If any of these attributes for the Customer table/dimension are not Primary Indexes, the updates will be very slow.

What I am trying to investigate is that in Teradata updates can be very slow if the Join condition is not based on Primary Index's. (AMP's and the parallelism withstanding). How to overcome this in practice?

	<Prev	Next>		<<First	<Prev	Next>	Last>>