Analyzing Consumption based on Workload

  • 1.  Analyzing Consumption based on Workload

    Posted 09-21-2020 14:23

    Workload behavior is an important component of your processing environment. Understanding when certain workloads are active and how this affects your overall consumption is crucial to understanding your z/OS system's behavior.  The z/Trim cost11 KPI plots the WLM-defined workloads against the R4H to provide insight into a workloads' contributions to the R4H and ultimately to your MLC costs. It is important to understand when your R4H peak is reached during a processing day and which workloads are the largest contributors.  This information is needed to make appropriate changes that can lower your MLC costs.

    If you are currently licensed with Sub-Capacity Pricing, the R4H is the strategic component of your Monthly License Charge. Even with TFP (Tailored Fit Pricing), understanding the workload consumption will give you valuable insights into your total consumption and resource utilization.  

    The cost11 screenshot shows the cost11 R4H Analysis by Workload Type for 09/08/2020. The R4H is shown throughout the 24-hour period. The R4H peak is 1,129 MSU's.

    The screenshot shows the R4H peak occurs at 12:00 PM. The R4H peak is almost all largely due to TP (online) processing. Particularly, the resource utilization from 9:00 AM through 12:00 PM accounts for the largest four hour consumption period of the day. The TP (online) afternoon processing only shows a three hour peak.

    This R4H Analysis by Workload Types shows that online tuning and balancing the teleprocessing traffic or activity throughout the day (showing most activity from 8:00 AM to 8:00 PM) may contribute the most to lowering the R4H peak and the MLC. Particularly assigning some work to run at 1:00 PM or 2:00 PM or after 6:00 PM may help to balance the online system consumption and lower the R4H. Optimizing the online systems in this example could also lower the R4H.

    WLM Report Classes allow you to distinguish between items that are grouped together in a z/OS Workload Manager service class. You can also combine transactions in different service classes together into one report class. This allows a company to build WLM Report Classes that can create an accountable entity than can be used to analyze, budget, prioritize and summarize resource consumption.    The Report Classes in our demo system basically complement the WLM Workload Types. The Report Classes are: Main Activities (largely featuring the batch workload), Transaction Activities (largely featuring CICS and DB2 workload) and Other Activities (some Batch, DB2 and OMVS work).

    The cost13 screenshot below shows the cost13 R4H Analysis by Report Class Family for 09/08/2020. It gives very similar results to the cost 11 screenshot.

    Once again, the majority of the resource consumption at the R4H Peak can be attributed to Transaction Activities. There are some minor differences between the resource assignments in the cost11 and cost13 KPI's, but they deliver a consistent message. In this environment, tuning should be focused on the online environment (both tuning and scheduling transaction activities) to lower the Monthly License Charge (MLC) based upon the R4H(Rolling 4 Hour Average).

    Mark Perillo
    Rocket Software