Archives of the TeradataForum

Message Posted: Thu, 06 Dec 2012 @ 20:51:00 GMT

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

As taught to me by Mike Buckland, the primary purpose of the PI for a normal Teradata table is:

1st) Distribution

2nd) Joining (especially between independent and dependant tables)

Of course, with a surrogate key as the index, #2 becomes difficult. With a flakey, serial number as the index, #2 is impossible. Are you propagating keys (yes, I said keys) through the logical model, so that maybe you can get good joining, if the key makes a good index? Look, if you are an heretical Teradata user and do things like Star schemas and such, Mike Buckland also emphasized that the purposes for indexes on those are even reversed!

1st) Joining

2nd) Distribution

Better know how a starship works (Star Trek: The Wrath of Kahn) before you go flying into that nebula.

iv

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