Archives of the TeradataForum

Message Posted: Wed, 20 Aug 2003 @ 21:41:51 GMT

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

If the client is cpu bound in the return phase on an SMP system it might help.

The throughput is limited by the slowest piece: Teradata or the client (including workload contention), the network from the client to Teradata, the I/O that FXP does.

Note that I've seen strange behavior with FastExport where the throughput declined dramatically with more sessions (10x). I would suggest running some baseline tests varying the sessions and of course ensuring 64K response buffers.

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