Managing shared resource pools for enterprise applications

The thesis presents a comprehensive management approach for shared resource pools that enables the efficient use of server resources while managing the resource access quality offered to workloads. The approach includes controllers that operate at three levels. A workload placement controller globally allocates and consolidates workloads based on historical demands. To improve resource access quality, a migration controller continuously observes the resource pool and migrates workloads between servers in order to decrease the duration of quality violations. Each server has a workload manager that locally optimizes resource allocation. A resource pool simulator models the placement of workloads, simulates the competition for resources, causes controllers to execute according to management policies, and dynamically adjusts placements and configurations. A case study shows that fully integrated controllers outperform the separate use of controllers or loose integrations.