Archives of the TeradataForum
Message Posted: Thu, 02 Mar 2007 @ 00:30:32 GMT
When a second query enters the system, how the resources will be shared will depend on the priority differences between the two queries and how much resource each is capable of consuming. If the are identical queries in terms of their characteristics, and they have the same priority, theoretically they should equally share the resources at that point. If they are the identical queries and the second has twice the priority of the first, then the first will be pushed down from 80% consumption of CPU to about 33%, with the higher priority query consuming about 66% (twice the first), assuming nothing else is active and each query is able to consume all the CPU it is offered.
This is theoretical because there are often other factors involved, locking, different levels of demand for resources, waiting for other resources, uneven consumption patterns. But when a higher priority new query enters the system, it gets its entitled share of CPU, if it can use it, and this happens immediately.
Prioritites are constantly being re-evaluated as new work enters the system. But any unused CPU that a higher priority query cannot consume is immediately made available to lower priority queries, so lower priority work often consumes more CPU than higher priority work.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|