Archives of the TeradataForum

Message Posted: Thu, 24 May 2001 @ 16:15:19 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Karen,

No, you should be ok (I've never had problems with restarts). As far as I'm aware, Multiload restarting works like this;

During the acquisition phase, Multiload writes the data to it's work table(s). This table basically has the same definition as the corresponding target table, but t here are a few extra columns. One of the extra columns is a source-sequence-number, which is the input record sequence number. At the end of the acquisition phase, Multiload does a sort of this work table to achieve a number of things, one of which is to eliminate duplicate rows - it uses source-seq-number to achieve this.

Assume that you're checkpointing every 10k records and you fail after 11,500 records. When multiload restarts, it will restart loading at record number 10,001, i.e. the work table will hold records 10,001 to 11,500 twice. However, the sort at the end of the acquisition phase will then eliminate these 'duplicates', which means that the target table only gets an input record loaded once.

I'm not sure about the checkpoint processing during the Application phase, but (again) I've never found any problems as a result of restarts and checkpointing.

Hope that helps,

Dave

	<Prev	Next>		<<First	<Prev	Next>	Last>>