Capacity Analytics

In our previous update for IBM Z Performance and Capacity Analytics in February, we detailed some important updates to aid capacity planning activities and expanding the scope of our performance insight capabilities for z/OS Communication Server users.  I’m pleased to share details of more new features that add further enhancements to support performance analysts and capacity planners by enabling workload simulation, toleration and exploitation support for customers adopting latest levels of z/OS Connect EE, and further opportunities to lower the overhead of processing SMF data to drive vital performance insights.

Simulation of workload transaction response time and batch throughput delays

An example of where you can take historical performance data and use it intelligently for capacity planning is in forecasting and simulation of workload growth. When your environment changes different workloads respond in different ways. Your online work may experience performance degradation as the volume of work increases. This can lead to missed SLAs and operational problems. Batch workloads could take longer to complete which in turn could have an impact on other workload if overrunning into peak business periods.

You may have known trends of how these workload increases year on year but other times the change may be unexpected. The ability to apply different levels of growth over the coming years and see at a workload level what the impact to performance will be and begin to focus in on areas of interest. Maybe a small amount of consistent growth can be easily accounted for, but perhaps above a certain level there needs to be more detailed planning to take place, either to re-balance workloads to maximize use of existing resources, or to start making plans for capital expenses.

Traditionally, capacity planners would be creating lots of complex models and in-house tooling to build out their planning. These can often be inflexible and require significant rework as assumptions and workload patterns change. Here we have a brand new feature within IBM Z Performance and Capacity Analytics that enable a workload simulation forecast. By selecting a single LPAR and an interval period, an analysis can take place on what the potential impact might be on transaction or batch throughput delays if the scale of workload was to change in coming years. The variables that can be applied include a workload growth (or decrease) factor and hypothetical hardware changes and upgrades. You can apply different factors to see a full range of changes that could occur and determine the level of risk that potentially is being taken on with no changes to the current configuration or when might be the optimal time to take proactive actions to avert performance problems.

The pre-defined reports allow the data to be consumed and analyzed in various forms. The breakdown of data can also be analyzed in terms of the impact on online transactional workloads that may be affected by the growth such as expected delays that impact SLAs and the elongation time of batch workloads to complete. Your performance database is a valuable resource to mine here and so leveraging this to make quick forecasts on workload performance should save time and resources in the long run.

Support for z/OS Connect EE SMF 123 version 2 records
In APAR PH33645 we have extended the existing support for SMF 123 records to tolerate and support the new version 2 of the record enabling customers exploiting the latest updates of z/OS Connect Enterprise Edition. Version 2 of these records contain more detailed information about about each request, including information about where the request was sent. The new record also contains data for multiple requests in a single record, making it more efficient as the server information is captured once per interval rather than for each request.


As well as updating the internal tables to store the new fields from the records, the existing predefined reports have been modified to help give a clearer view on API and service level information. Using these reports you are able to quickly identify performance changes within z/OS Connect instances by looking at the trends on response time and transaction volume. The easy to adjust settings on the report allow you drill down to varying levels of granularity, including SMF interval level, and observe performance details on individual API/service over a defined period of time.

zIIP offload enablement of batch data collector
IBM Z Performance and Capacity Analytics added zIIP offload exploitation to the continuous collector process with original release of V3.1 in 2020. This allowed user making use of the “hub and spoke” architecture to further lower the overhead of processing SMF records in near real-time allowing access to reports and insights within minutes of the SMF records being created. Now we are pleased to extend the capabilities to user making use of the batch data collector to load and process large volumes of data in a single update through APAR PH33033. This enables more flexibility to make decisions on when to batch load data because you know the impact on general CPs will be reduced.

Where to learn more
If these updates are of interest to you and you’d like to learn more about IBM Z Performance and Capacity Analytics please do reach out to me and I’ll be happy to answer your questions. We have an on-demand webcast that will give you a full overview of the product so you can learn how it can help you make better use of your operational data for performance analysis, capacity forecasting and cost management. There are also some hands-on virtual lab session scheduled for June 9 and June 17 to give you guided experience in generating and viewing reports in the modern user interface.

Originally published on the IBM Community Blog.

Leave a Reply

Your email address will not be published. Required fields are marked *