A resource provisioning model for virtual machine controller based on neuro-fuzzy system

Resource usage overloading of virtual machines can occur in a virtualized data center when load changes dynamically over time. To control load unbalancing conditions in the virtualized data center, three-level of controllers are proposed. Among these three level controllers, we emphasize virtual machine level controllers (VLCs) on this paper. VLC can automate the task of monitoring, demanding required resource capacity from the host they are existed on. This paper presents a resource provisioning model for VLC based on neuro-fuzzy system for supporting CPU utilization decisions of VLCs. Workload - resource mapping and resource - resource mapping are conducted to predict resource usages of virtual machines. The model simulates the resource provisioning model and the results are evaluated for training and testing phases.