 |
 |
Archives of the TeradataForum
Message Posted: Thu, 02 Feb 2006 @ 10:45:27 GMT
| Subj: | | Re: Use of Multiset for Large Table |
| |
| From: | | Victor Sokovin |
| | Just re-read Michael's post: | |
| | There's no need for a Dup Check, because the PI is already unique and > if some columns are unique then all columns must be unique, too
:-) | |
Dieter, I know *that*. Just re-read the context. We had a SET table and row uniqueness was provided by SET (it is unlikely that such a table
had UPI on it, right?). Now, we change it to MULTISET but still want to guarantee row uniqueness. So what do we consider? New UPI, new USI, new
ETL logic, etc. And than insert my $0.02 comments. I hope this clarifies the context. If not, we should talk it through in a nice Biergarten when
the weather permits!
I did agree with what Michael said, BTW. He was right that it was a strange idea to consider UPI here. I just wanted to explain why one would
look at that idea at all. I myself do not think that UPI or USI should be considered; they are not practical options. The only alternative to SET
I can see is the logic in ETL plus independent checks in QA.
Regards,
Victor
| |
