Archives of the TeradataForum

Message Posted: Fri, 22 Feb 2002 @ 19:36:55 GMT

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

Hi All,

I'm looking for some advice on PI index choices.

Because of business rules, I need to join a 1.5 million row (current UPI is col2) table to a 1.5 billion row table (current UPI is col1, col2, col3) as well as to a calendar table (UPI col1).

Performance would be greatly improved if I modified the large table to have a NUPI of col2, I'd be doing PI joins to the two tables. However the NUPI rows/value would be about 1000. There would be fairly even distribution across amps with each value have close to the same number of rows/value (no spiking)

Questions are:

1) Anybody had to implement this kind of scenario?

2) would putting a USI (that would hardly ever be used) just to enforce uniqueness be worth the DASD and processing overhead to eliminate the NUPI duplicate row check?

Ruth

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