Enabling Workload Interaction Correlator

The following topics describe how to enable IBM z/OS Workload Interaction Correlator.

IBM® z/OS® Workload Interaction Correlator adds or updates the following parameters in the SMFPRMxx member of parmlib:
- HFTSINTVL (updated)
- WIC (new)
These are further described in HFTSINTVL — Specifying the high-frequency throughput statistics interval and WIC — Specifying the generation of type 98 records for IBM z/OS Workload Interaction Correlator.
For complete information about the SMFPRMxx parameters, see Statements and parameters for SMFPRMxx in z/OS MVS Initialization and Tuning Reference and Customizing SMF in z/OS MVS System Management Facilities (SMF).
To enable IBM z/OS Workload Interaction Correlator, complete the tasks in IBM z/OS Workload Interaction Correlator.
For information about the applicable changes to the SMF type 98 record, see Record type 98 (X'62') — Workload interaction correlator and high-frequency throughput statistics.

IBM z/OS Workload Interaction Correlator

You can configure SMF to collect IBM z/OS Workload Interaction Correlator data. IBM z/OS Workload Interaction Correlator enables a product, such as a z/OS component, middleware, or application, that executes in multiple instances distributed across multiple address spaces to generate high-frequency, summarized, system-wide and compartmental data. Products can use the IBM z/OS Workload Interaction Correlator services to generate an SMF type 98 record for a product's subtype every 5 seconds containing data about the product's activities in a standardized, synchronized, contextualized format.

Records from the workload interaction correlator produce a summary view of the component's activities across address spaces of similar priority and CPU usage and highlight data from exceptional address spaces. These records are generated on a synchronized, high-frequency 5-second interval across multiple products.

Analytics tools, such as IBM z/OS Workload Interaction Navigator, can analyze these records across z/OS components, middleware, and applications to detect and correlate anomalies. Presenting an organized progression of interacting anomalies reduces the time and expertise needed to diagnose a performance problem. When an action is taken to solve or mitigate the performance problem, the system programmer can verify that the fix is behaving as expected by comparing the results before and after the implementation of the action.