CPU utilization

As a rule, the resource CPU is used simultaneously by many tasks. Here, too, the size of the workload is a key criterion for the amount of time a task has to wait for the CPU.

The average hardware service time of the resource CPU per user request (e.g. between two DMS I/O operations) is relatively small as compared to the execution of I/O operations.

Consequently, the wait time in front of the resource CPU has considerably less effect on response or dwell time than the wait time in front of disk drives (volumes). Therefore, a heavier workload is feasible as compared to shareable volumes. However, for the main application a workload of 70% should not be exceeded (with multiprocessor systems - depending on the number of CPUs - a maximum workload of 90% can be tolerated). When multiple guest systems are used, generally a 5 - 15% higher overall workload than in native operation is possible.

In addition, the task scheduler PRIOR allows tasks to be given precedence by means of priority assignment. This makes it possible to use part of the remaining 30% (or 10%) by running low-priority tasks (with correspondingly long wait periods in front of the resource CPU). Utilization of 100% of the CPU without adversely affecting the main application will only be possible in rare instances.

ANALYZER reports for assessing the CPU workload

Report group	Report	Meaning
CATEGORY- CPU	CPU utilization (TU+TPR) for category	CPU workload in the TU and TPR processor states for each task category
CPU	Utilization real	CPU workload in the TU, TPR and SHI processor states (actual values). When VM2000 is being used, this report is important for determining the CPU share of the guest system vis-à-vis the overall system
	Active logical machines	Number of active logical machines (number of CPUs used by BS2000)
	Sum SVC calls	Sum of all SVC calls in the TU and TPR processor states