Archives of the TeradataForum

Message Posted: Fri, 28 Jan 2005 @ 16:54:11 GMT

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

<-- Anonymously Posted: Friday, January 28, 2005 10:52 -->

Chris,

I am afraid that looking at the number of users on the system can be extremely misleading because of the complexity of the system and the variation in users (not all users are equal in terms of activity/utilization).

System utilization with the application mix, the number of concurrent utilities, transient journal, changes to ddl, locking methodologies, the size of the access rights table, statistics, ....

Standard queuing theory suggests you need to know the average wait time for a process in the 'queue'. Standard queuing theory does not readily apply to teradata because of the number of nodes (and CPUs) that have to work in concert with each other. Experience would suggest that knowing the average number of active users helps but it really takes a lot of work to predict capacity for any given system...

Hope this helps

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