Archives of the TeradataForum
Message Posted: Wed, 09 May 2001 @ 13:37:08 GMT
Subj: | | Re: Unique Row ID |
|
From: | | Eric J. Kohut |
Claybourne is correct; However, I would just use columns separated by a comma and not concatenated together. The effect is the same
without converting all of the columns to character first. As long as the columns values in total are Unique then the CSUM will create a
unique key id. Be careful of using columns with nulls due to the unknown nature of the value.
Also, you can cast the original number 1 as 1 (decimal(18,0)) this might help you get past the integer limitation if that is close.
While CSUM is a little resource intensive, it should be the fastest way of creating this unique key id. You shouldn't have much of a skew
problem if the values in the column list are unique. Spool of course is always a problem on your entire largest table in the system. Make
sure you have enough.
Good Luck,
Eric
EJK
Eric J. Kohut
Senior Solutions Consultant - Teradata Solutions Group - Retail
Certified Teradata Master
NCR Corp.
|