 |
 |
Archives of the TeradataForum
Message Posted: Sat, 12 Apr 2003 @ 05:52:30 GMT
Subj: | | Re: How to implement the sequence concept in teradata |
|
From: | | Walter, Todd A |
See the Identity column feature in V2R5. As previously noted, this does not guarantee a monotonically increasing sequence but it does
guarantee a unique key to identify the row. It works with bulk SQL (eg INSERT SELECT) and single row SQL (eg INSERT VALUES).
Re identifying the subscriber by first, last and address: There are really good reasons to have a "subscriber id" generated in the
database or in an application outside. Without a subscriber id, it would appear that there is a data modeling problem (eg is it really a
different subscriber if the name or address changes?). Typically a subscriber id is assigned long before the data makes it to the warehouse
but if this is not the case, then it is a good reason to use Identity column.
| |