CPU Usage Prediction Models for Virtualized Data Center

Resource allocation plays an important role inVirtualized Data Center (VDC). The applicationsrunning in VDC are mostly business criticalapplications with Quality-of-Service (QoS)requirements. Moreover, dynamic resource allocationand real time monitoring of the resource usage of VMsare also needed to reduce under resource utilization andover resource utilization. Therefore, resource usageprediction is required for dynamic resource allocationsystems. In efficient dynamic resource allocation, theresources are allocated to a VM while meeting theirService Level Agreement (SLA). The main contributionof this work is two-fold. The first is the generation ofCPU usage prediction models by applying differentpowerful machine learning techniques. The second isSLA evaluation on predicted value by using proposedSLA metric. To evaluate the efficiency of these models,experiments are carried out by using CPU profiles fromreal world data centre. According to the experiments,proposed resource prediction models have promisingaccuracy