Often, slow moving growth or declines may go unnoticed if you are looking at overall demand. It becomes easier to miss any changes and trends unless there is an ability to zero in on more granular buckets of usage. What’s needed is a capability to start at a high level, and then be able to parse out pieces of the workload based on different dimensions of work.  

Sometimes a few product lines or workload types might be enough to see the high-level trend, but correlating with business changes may require more granularity than is provided by service classes. Also, service classes are oriented toward workload prioritization, not toward understanding business costs. Thousands of report classes may help identify some segmented details of workload type and business related work, but it may prove difficult to see the forest through the trees.  

Many sites have turned to external solutions to navigate the workload characterization maze. Marrying business activity with workload consumption is the main purpose behind workload characterization activity. Whether it’s insurance claims, packages delivered, passengers ticketed, a meaningful translation of resource consumption and costs into business activity is often filled with dead ends and tribal secrets.  

I’ll explore one use case that may help motivate your efforts to improve the line of sight between resource utilization and business activity. While business metrics are important, this blog is focused on defining applications as a starting point. 

What is the Driver?  

Capacity planning inevitably meets with the CIO directly or indirectly to request funding for upgrades. It doesn’t matter if it’s CPU, disk, tape, or network equipment; the questions always include something like: “What is driving the need for 10% more resources, when the business is growing just at 4%?”  

Sometimes the relationship between consumption and business revenue is non-linear. There better be a pretty amazing margin on the revenue to make up for that significant increase in resources required. Perhaps, a mismatch between revenue and resource consumption is ok for a small segment in a test market situation, but it may present a business risk at scale. Regardless, this should help you see the need for segmenting workloads.  

After reviewing the growth plans and the key business segments, the CIO has some good answers to provide the board for the next 10% hike in IT infrastructure costs. Let’s look closer at this phenomenon within some data. 

Drilling Down into the Details 

The report below from IBM Z IntelliMagic Vision for z/OS contains the tailored fit pricing (TFP) metric, weekly MSU consumption. There’s approximately a moderate ~10% growth over the course of two years.  

Figure 1: Weekly CP Dispatch Time (MSU Consumption)  

One way to identify the growth is shown in Figure 2. There, we begin to see that the growth is primarily in the Batch & CICS workloads, while we observe the System work has actually decreased.  

Figure 2: Weekly MSU Consumption by Workload – Closing in on the Growth

Further Detail 

In our use case, the customer applied tags on many different segments of the infrastructure consumption to define many separate applications automatically as the SMF data is processed. This categorization can then be used to ‘peel apart the onion’ and segment the workloads into helpful business related categories that will help to correlate resource changes with business changes.  

It is common among mainframe sites to observe hundreds of different applications running on the mainframe. Any given application might include batch and online workloads, and those workload types often rely on different subsystems and software stacks within the infrastructure. In fact, the major increases in Figure 2 indicate the CICS and batch workloads drove more than the 10% growth.  

Let’s assume the increases in Batch and CICS workloads are understood to be driven by transaction growth for the moment. That might be another blog post at a later date. It turns out there’s some interesting things in the decline in the systems workload. Figure 3 takes a view of one of the applications defined that contains some System work, Monitor.  

Figure 3: Monitor MSU Consumption (Tailored Fit Pricing) & Average MSU Consumption 

In Figure 4, we can begin to see what’s underneath the Monitor application as it’s comprised of 4 report classes. 

 

Figure 4: Monitor report class details 

This report shows some pretty clear breaks in consumption. RTMON_STC shows a pretty steady decline with a bigger break in usage around the end of June 2023. The second largest contributor in 2022, RTMON, saw one drop around September 2022, and another marked drop just before March, 2023.  

Tying those changes to business actions is the next key. What we do know is that the RTMON_STC drop in mid-2023 coincided with an outsourcing change for this customer. This event likely resulted in a switch in usage patterns. It’s not a stretch to see how that might lead to less monitoring or that new techniques require less polling type monitoring as other solutions are employed.  

Uncovering the Details Enables Collaboration 

IBM Z IntelliMagic Vision for z/OS provides drilldown capabilities to address space detail when identifying changes. This enables IT applications, infrastructure and middleware owners to effectively communicate needs to business lines and product lines. An example of this is provided in Figure 5.  

Figure 5: Address space level detail

Sorting them from highest to lowest with a push of a button on screen can help you quickly identify the root cause of the decline. This example helps show the principle behind assignment of the work at a granular level into report classes, and then aggregation of several report classes into applications with attributes of workload types, systems, sysplexes and more.  

The value in a solution that attributes resources to applications, products, and other business and infrastructure related characteristics will help business and IT become more effective as the partners deploying IT resources to meet the needs of their customers.  

If you’re looking for ways to improve workload visibility and correlate CPU consumption with business activity, the webinar "Integrating Mainframe Resource Consumption into the Business" covers key strategies to bridge the gap between IT resource usage and business outcomes.

We would welcome an opportunity to introduce IntelliMagic as an effective solution to help you navigate the maze and deliver clear insights into the demands on z/OS resources.