Archives of the TeradataForum
Message Posted: Mon, 05 Jul 2004 @ 14:35:30 GMT
| Subj: || || Re: Order of Records in Source file for MLOAD |
| From: || || Victor Sokovin |
| ||If a source file has more than one record with same primary index (unique) value, what happens when an UPSERT MLOAD is run to load this
file into a teradata table (with unique primary index) ?.|| |
| ||Will the first record (as per the sequence in the file) be INSERTED (assuming the table did not have the record before) and will the second
record update the just INSERTED record ? If yes, where does this UPDATE happen; in the WORK table before applying to target table or while
applying to the target table ?|| |
ML executes UPDATE and INSERT (in this order) in the work table so I think it won't meet your expectations in this case. The first record
in your example will be inserted and the second record will be treated as a duplicate. What happens next will depend on the instructions regarding
duplicates that you have given to ML.
In my opinion, the best way to proceed is to run a few tests with all possible cases.