Archives of the TeradataForum

Message Posted: Thu, 01 Jan 2004 @ 07:48:13 GMT

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

I strictly object this, it is surely in seqential fashion if the batch size is 1 and is incremented by 1. If not so, please explain the practial scenario with example, may be I can correct myself!

When it is a batch size of 1, each amp while inserting the row checks with the dbc.idcol table and gets the latest value for that particular table and updates. This value doesn't reside at the amp itself. So even if an amp goes down the sequence is still maintained and hence the data load is strictly sequential.

Also have a look at the below note of the developer of Identity column to a customer having this same issue,

"If Identity Column usage is solely restricted to assignment of contiguous unique identifiers (e.g. Client ID in Hartford's case), the DBSControl field, IdColBatchSize, may be set to 1 to ensure tha numbering is sequential."

So, I would well suggest Shabahat to go ahead with a batch size 1 and increment by 1, if not much concerned about performance.

-Prasath.

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