A Stable Matching-based Virtual Machine Allocation Mechanism for Cloud Data Centers

Virtualization is the enabling technology that makes resource provisioning in Cloud computing feasible. With virtualization, virtual machines (VMs) can be migrated across physical hosts to achieve better utilization of resource with a minimum impact on service quality. The VM allocation problem can be formulated as a stable matching problem. In this paper, we propose a VM allocation mechanism based on stable matching. A deferred acceptance procedure is adopted to handle conflicts among preferences of VMs and physical hosts. Unlike ordinary stable matching problems, both involving party groups in our matching process are having a mutual objective, that is to reduce the overall energy consumption of a Cloud data center while maintaining a high level of Quality of Service. The proposed mechanism is evaluated using CloudSim with real-world workload data. Simulation results show that Cloud data centers with the proposed mechanism can reduce energy consumption and avoid violations of ServiceLevel Agreement.