  |  
  |  
 
Archives of the TeradataForum 
 
 
Message Posted: Wed, 17 Oct 2007 @ 21:41:00 GMT 
 
  
 
 
 
 
  
|  Subj:  |   |  Re: Removing duplicates from table  |   
|     |   
|  From:  |   |  frank.c.martinez  |   
  
 
 
  
Ah, another Partners moment.  Yes, they talked about have INSERT/SELECT use error tables in a future version of Teradata.  Wouldn't that be
wonderful?  But fer now, you're outa luck, pilgrim.  Remember that "first" doesn't necessarily mean anything in a parallel environment.  Today,
AMP X on NODE Y returns a record first because nothing was happening on it, whereas tomorrow (or even five minutes from now) it could be AMP Z on
NODE W.  The reason "first" means anything in a fastload is because the rows are in some sort of order in the incoming file, and since we're
talking a UPI collision, the two records are going to the same AMP.  Now again, I ask the perennial question: 
"What in the (business) are you trying to accomplish?" 
 iv 
 
 
 
 
   
 
 |   |