Fair Share

Enhanced Definition

In the context of IBM z/OS, **Fair Share** refers to a Workload Manager (WLM) principle that ensures equitable distribution of system resources, primarily CPU, among competing workloads. Its primary purpose is to prevent any single workload or group of workloads from monopolizing resources, thereby maintaining system responsiveness and meeting defined performance goals for all critical applications.

Key Characteristics

- WLM-Driven: Fair Share is a fundamental concept implemented and enforced by the z/OS Workload Manager (WLM) based on the active WLM policy and defined service definitions.
- Resource Equity: It aims to provide a proportionate share of available resources (CPU, I/O, memory) to different workloads, especially when the system is constrained or experiencing contention.
- Goal-Oriented: While ensuring equity, Fair Share operates within the framework of WLM goals, striving to meet the performance objectives defined for each service class.
- Dynamic Adjustment: WLM dynamically adjusts resource dispatching priorities and allocations in real-time to achieve Fair Share, responding to changing system conditions and workload demands.
- Configurable Granularity: Fair Share can be applied at various levels, such as within a service class, across multiple service classes, or to specific resource groups, allowing fine-grained control.
- Prevents Starvation: It is crucial for preventing lower-priority but still important workloads from being completely starved of resources by higher-priority, resource-intensive tasks.

Use Cases

- Batch Job Processing: Ensuring that multiple concurrent batch jobs, especially those from different business units, receive a fair share of CPU cycles, preventing a single long-running job from significantly delaying others.
- Mixed Workload Environments: Managing resource allocation between online transaction processing (CICS, IMS), batch processing, and started tasks within the same LPAR to maintain responsiveness for interactive users while still progressing batch work.
- Departmental Resource Management: Allocating system resources fairly among different departments or applications (e.g., finance, HR, inventory) that share a common z/OS system, based on their defined service level agreements.
- Peak Load Management: During periods of high system utilization, Fair Share helps WLM distribute the available resources equitably, preventing a complete collapse of service for any critical application.

Related Concepts

Fair Share is intrinsically linked to the Workload Manager (WLM), which is the z/OS component responsible for managing system resources and enforcing performance goals. WLM uses the Fair Share principle when dispatching work units within service classes and report classes, ensuring that workloads receive their proportionate share of resources based on their importance and defined goals. It operates within the context of an LPAR (Logical Partition), distributing resources among the workloads active within that LPAR. Fair Share complements other WLM concepts like importance and goals by providing a mechanism for equitable resource distribution when multiple workloads are competing for the same resources.

Best Practices:

Define Clear WLM Goals: Establish realistic and measurable WLM goals for all critical workloads to give WLM the necessary targets for applying Fair Share effectively.
Monitor WLM Performance: Regularly use tools like RMF (Resource Measurement Facility) and SMF (System Management Facilities) to monitor WLM's effectiveness in achieving Fair Share and meeting goals.
Review Service Definitions: Periodically review and tune WLM service definitions, especially service classes and resource groups, to ensure they accurately reflect current business priorities and resource requirements.
Understand Importance Levels: Be aware of how WLM **