Archives of the TeradataForum

Message Posted: Tue, 28 Jun 2005 @ 20:20:53 GMT

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

Swapnil -

If I understood your question correctly , you can't achieve both by selecting different PI on parent and child ..We all know that . To achieve that kind of scenario you need to go with JOIN INDEX but you said you have space constraint . I just want to know how skewed the data on child table with PI as ACCOUNT_ID ? If you have test data on the physical tables try to find out the no: of collisions for row hash on child table ...if its not more than 10 then I don't see any problem in making Account_id as NUPI on Child table ...so UPI vs NUPI join is always 1 AMP operation

Thanks,

Srini

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