Chapter Overview

The 1909 Ford Model T is widely known as the first automobile produced on an assembly line. The vehicle had a wooden dashboard with brass trim, leather upholstery, and a speedometer. Some models were even equipped with windshields. Since the initial introduction of the Model T, countless safety and convenience features have been added to automobiles. Technologies such as intermittent wipers and cruise control are now standard features in most automobiles. Intermittent wipers allow a driver to focus on the road instead of having to repeatedly turn the wipers on and off. Similarly, cruise control gives drivers the ability to set the speed of the vehicle and focus on driving instead of having to consciously keep the vehicle at a fixed speed.

The Workload Manager product in HP's Virtual Server Environment offers advances in computing technologies analogous to the features that have been added to the original Model T. Figuratively speaking, WLM adds intermittent wipers, cruise control, and other automation capabilities to computing environments. Like intermittent wipers, Workload Manager provides automated resource allocation so that administrators don't need to monitor workloads and perform manual adjustments. Like cruise control, Workload Manager allows system administrators to configure the desired response time and adjusts the throttle to ensure that resources are provided to the workload to maintain the desired response time.

This chapter begins with an architectural overview of the Workload Manager (WLM) product and its capabilities. The fundamental terminology regarding Workload Manager is covered next. Then a detailed scenario involving the integration of Workload Manager and virtual partitions (vPars) is described. In the scenario, Workload Manager is configured to migrate CPUs between two virtual partitions based on the resource utilization within each vPar. A second example scenario is covered that demonstrates Workload Manger's integration with Serviceguard. In the second scenario, Workload Manager is utilized in a Serviceguard failover environment to ensure that a workload that has failed over to an adoptive node receives the resources necessary to sustain the workload. In situations where the adoptive node does not have adequate CPU resources, Workload Manager activates unlicensed Temporary Instant Capacity CPUs to ensure that the production workload meets its service-level objectives (SLOs).