Archives of the TeradataForum
Message Posted: Tue, 18 Jan 2005 @ 09:13:26 GMT
| Subj: || || Re: FastLoad VS MLoad |
| From: || || Victor Sokovin |
| ||If your data were to be badly skewed(i.e. considering NUPI) and you have a SET table to be loaded then MLOAD takes a longer time - because
it checks every row to be inserted for duplicates. Added to this error logging (logging the duplicate row to the error table) would again take
some cosiderable time. You could avoid this if you were to have a MULTISET table as the target table to be loaded.|| |
If MULTISET is not an option, IGNORE DUPLICATE ROWS in the MultiLoad script might help make the competition a bit more fair. The target
table should not have a UPI, though. Otherwise, specifying this clause won't help as it will be ignored.
I still wonder, however, whether this will explain the whole run time difference. One extra hour for 60 mln rows is a lot.