A Resource Pool Management Model using Fuzzy Logic Decision Making

In Virtualized Data Center (VDC), fixed allocation of virtual machines (VMs)' resources usually make overloading or under-loading condition as the workloads of the application they run have dynamic features. Underloading leads unutilized resource usages and SLA deviation is also appeared because of the over-loading. Therefore, dynamic allocation of VM resource usage is necessary in order to get a load balanced environment. Resource usage prediction of virtual machines is needed to be done for dynamic VM resource provisioning. For VMs, dynamic resource provisioning is available only when the host has enough re-sources to complement their demands. But in the condition of host overloading, VM migration should be made in order to keep a load balanced condition. VM migration depends on a variety of criteria and efficient decision support is required. In this paper, neuro-fuzzy is applied for resource usage prediction and combination of AHP and Fuzzy-TOPSIS is used for decision support of VM migration. VM resource prediction and decision support model are evaluated by the resource pools data. Simulation and mathematical evaluation confirms that such an integrated solution ensures acceptable accuracy.