z/VM Performance Data Pump

New function is provided in z/VM Performance Toolkit to integrate z/VM performance data in an existing Splunk observability platform to populate real-time dashboards and generate reports.

News

• 2025-05
  The z/VM Neural Network Processor dashboard is available from the Grafana Labs Dashboards site with ID 23458.
  On z16 the dashboard shows the AI Accelerator usage by topology; on z17 the sharing of AI Accelerators is also reported. The metrics are provided by Data Pump when the CPUMF data collection is enabled and appropriate service has been installed. For z/VM 7.4 APAR VM66826 is required, for z/VM 7.3 function is provided by APAR VM66825.
  A detailed description of the dashboard is provided as z/VM Neural Network Processor.
• 2025-04
  A new sample dashboard z/VM Energy Consumption is available from the Grafana Labs Dashboards site with ID 23265.
  The power usage metrics are provided by z/VM when running on z17 hardware and appropriate service has been installed. For z/VM 7.4 the APARs VM66824 and VM66826 apply, for z/VM 7.3 the APARs are VM66823 and VM66825. When provided by the z/VM monitor, the metrics are extracted with the default configuration of the z/VM Performance Data Pump.
  A detailed description of the dashboard is provided as z/VM Energy Consumption Dashboard.
• 2025-03
  New revisions of the sample dashboards have been published on the Grafana Labs Dashboards site to correct some missing data when using Grafana 11.

Additional Resources

• For an introduction into Performance Data Pump, see IBM z/VM Spotlight and Performance Data Pump Overview and Demo.
• For customers who prefer an open-source solution, data can be sent to the InfluxDB time-series database where performance data from different sources can be retained and visualized using Grafana real-time dashboards. Sample dashboards for Grafana are available. Customers who wish to deploy InfluxDB and Grafana in a Linux guest on z/VM can use the container images available in the IBM Z Container Registry. Review Grafana Dashboards - Fast Start Guide that covers all the steps from installing the PTF to deploying containers in a Linux guest on z/VM and any aspects to consider when moving beyond a proof of concept implementation.
• The z/VM Performance Data Pump is pre-installed with z/VM 7.4 as user DATAPUMP. For z/VM 7.3, APAR VM66687 delivers the new function. PTF UM90334 for z/VM 7.3 can be ordered.
• Sample dashboards can be found on the Grafana Labs Dashboards web pages by searching for z/VM.
• Appendix I is added to "z/VM: Performance" (SC24-6301-73). The "z/VM: Other Components Messages and Codes" publication (GC24-6300-73) is updated to include the Data Pump messages. HELP files are also provided for Data Pump configuration with HELP FCX DATAPUMP.

Frequently Asked Questions

• Why are my Linux guests not reported as "Linux" workload on the dashboard?

  The information is based on information provided by the guest operating system to the z/VM hypervisor. Some older Linux systems don't do that yet. Current RHEL and SLES releases provide that information. Ubuntu 20.04 introduced the change to provide this information.

• Why are my service virtual machines to classified as such?

  The service virtual machines are recognized by the OPTION SVMSTAT directory statement. Most standard service virtual machines installed with z/VM are marked as such. If you have additional virtual machines that you want to see distinguished from the "Other" class of workload, you can add the statement to the directory entry.

• Why are z/VSE guests reported as "Other" workload?

  The current z/VSE level does not use the hypervisor interface to identify the type of workload. We're working with the vendor to get that support added at some point.