Work Load Manager

Strobe reports on how your application programs use resources in IBM z/OS environments. It determines where and how an online or batch application consumes and waits for CPU time, or waits for I/O to complete. However, the measured address space is just one element of the overall workload competing for system resources at any given moment. IBM's Workload Manager is a solution for managing the distribution and balancing of workloads, and the allocation of resources to these competing workloads. The object is to meet performance goals assigned importance based on business needs. Workload Manager is dynamic in monitoring workloads and adapts processing based on the defined business goals.

iStrobe's Work Load Manager report shows the active Workload Manager environment, definition, and policy during the measurement. The report also shows the workload and service class that the measured address space is assigned to.

WLM Environment

  • Functionality: Indicates the highest level of WLM functions that exist in this service definition.

  • Version: The version or release of WLM.

Work Load Definition

  • Name: The identifier of a collection of service classes which in turn represent a collection of work; e.g. CICS, IMS, production, test, batch, online, etc.

  • Description: 32 characters of text describing the work load of which the current service class is associated.

Service Definition

  • Name: The identifier of the service definition containing all of the data necessary for workload management processing.

  • Description: 32 characters of text describing the service definition.

  • Date Installed: The date the service definition was installed on the WLM couple data set.

  • Time Installed: The time the service definition was installed on the WLM couple data set.

  • Installing User ID: The user identifier of the service administrator who installed the service definition on the WLM couple data set.

  • System ID Installed From: The system name on which the definition was installed.

  • CPU Coefficient: The additional weight assigned to CPU service.

  • I/O Coefficient: The additional weight assigned to I/O service.

  • Storage Coefficient: The additional weight assigned to storage (central page frames) service.

  • SRB Coefficient: The additional weight assigned to service request block execution time.

Service Class Definition

  • Name: The identifier of the active service class naming a group of work with similar performance characteristics; performance goals, required resources, and business importance.

  • Description: 32 characters of text describing the active service class which can be associated with only one work load.

  • Number of Periods: There can be up to 8 periods per service class because some work may have varying resource requirements.  Each period specifies a goal and importance.

  • Current Period: A numeric value from 1 to 8.

  • CPU Critical: An indicator regarding CPU protection.  If Y(es) then other work of lower importance always has a lower dispatch priority.

  • Storage Protection: An indicator regarding storage protection.  If Y(es) then this work will lose storage to other work only if the other work is of equal or greater importance, and needs storage to meet performance goals.

  • Used in Transaction Subsystem: An indicator regarding service class usage.  If Y(es) then this class is used in a transaction subsystem type.

  • Used in Address Space Subsystem: An indicator regarding service class usage.  If Y(es) then this class is used in an address space subsystem type.

  • Used in Enclave Subsystem: An indicator regarding service class usage.  If Y(es) then this class is used in an enclave subsystem type.

  • Used in Non-MVS LPAR: An indicator regarding service class usage.  If Y(es) then this class is used in a non-z/OS partition in the LPAR cluster; e.g. Linux.

  • Discovered Time: Since an address space can be reassigned to different service classes, this represents the time that this service class was returned by the SYSEVENT call.

Service Policy

  • Name: The identifier of a service policy containing service class and resource group overrides, which are merged into service class and resource group specifications in the service definition at activation.

  • Description: 32 characters of text describing an active service policy.

  • Date Activated: The date the service policy was activated on the measured system.

  • Time Activated: The time the service policy was activated on the measured system.

  • Activating User ID: The service administrator or system operator user identifier activating this service policy.

  • Activated From System ID: Name of the system upon which policy activation occurred.

Period Definition

  • Type of Goal: The type of goal this period represents. There can be only 1 goal per period. Goal types are average response time, response time with percentile, velocity, and discretionary.  Response time is the amount of time expected to complete the work.  Velocity is a measure of how fast work should run when ready (without delay) for WLM-managed resources. Discretionary is defined by Workload management, run when excess resources become available.

  • Importance Level: The relative importance of the goal, used only when work is not meeting the goal.  Defined from 1 to 5 with 1 being the highest level.

  • Response Time Unit: The unit in which a response time goal is expressed.  Units can be expressed in one of milliseconds, seconds, minutes, or hours.

  • Goal Percentile Value: If the period has a percentile response time goal, the percentile of work in the period that should complete within the response time.

  • Duration: The length of the period in service units.  If the work included in this period does not complete when this number has been reached, the work moves to the next performance period.  The last period would contain 0.

  • Response Time Goal: The amount of time expressed by response time unit for response time and velocity goals. Zero for discretionary or no goal defined.

  • Velocity Goal: The expected amount of time, expressed as a percentage (from 1 to 99), of how fast work should run when ready.

  • Discovered Time: As resources are consumed, WLM assigns the address space to higher periods based on the class period definitions. This represents the time that this period was returned by the SYSEVENT call.