Archives of the TeradataForum

Message Posted: Mon, 10 Feb 2003 @ 15:14:30 GMT

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

Because Teradata reserves a "batch" of identity values at a time and caches them (on each AMP for INSERT/SELECT) for better parallelism. The "INCREMENTING BY" clause only applies within a batch. You have to make a tradeoff between "gaps" in the numbers and insert performance.

The batch size is controlled by a dbscontrol "general" field (IdCol Batch Size), which defaults to 100,000. From your sample, I'd guess your system has 20 AMPs and dbscontrol is set to the max of 1,000,000.

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