Archives of the TeradataForum
Message Posted: Wed, 21 Jan 2004 @ 13:34:52 GMT
A JI should never be more restrictive than the underlying base table(s). If it were, you could potentially get an error that a uniqueness rule was violated while loading data into a JI; however, the same data loaded into the respective base table(s) without any problems. This scenario could cause some major headaches.
So, maybe, to avoid this problem, my guess is that JIs were designed never to have UPIs. Performance shouldn't be an issue, because, whether or not a PI is unique, it will hash the same. So if I build a JI on a base table with the same PI (except the JI's PI is a NUPI instead of a UPI), joins back to the base table will still be AMP local because the 2 objects (the base table and JI) have the same PI; therefore, the data is hashed to the same amps.
I would really like to know the real answer, hopefully someone from Teradata can pipe in.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|