In this paper, we present a resource provisioning technique which classifies data center entities through their attributes as characterized by their resource requirements. The classification process enables the multi-level matching approach which assigns client requests to the most appropriate VM, while the VM is assigned to a Host of similar characteristics. By considering the multi-dimensionality of compute resources, data center entities which include applications, VMs, and Hosts, are consolidated and assigned according to the most proportionate available resource. Evaluation results confirm the benefits of applying such techniques on the performance and energy consumption of a cloud data center.
展开▼