Archives of the TeradataForum

Message Posted: Tue, 15 Mar 2005 @ 13:33:25 GMT

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

Bill,

I came across this some time back, but unfortunately cannot recall the best solution.

The slow performance is down to the blocking.

You are using a pack factor of 50, which means that an MSR will include 50 update statements.

Remember that row level locking is not, it is row hash locking and therefore it is possible 2 updates may require the same lock even with different PI values.

I believe the blocking is in effect caused by a 'conflict' between the number of sessions and the number of statements in the MSR (locks will not be release until the last update in the MSR completes).

I suggest you try varying the pack factor and number of sessions, and also consider the serialise option already mentioned. Sorry I cannot reclal the answer!

Regards

Keith.

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