Evolutionary algorithms based virtual machine consolidation and utilization prediction for energy efficient cloud data centers

Abstract

Increasing Cloud computing infrastructures have resulted in notable newlineenergy usage in cloud data centres. This demand for excessive energy not newlineonly results in significant operating costs, but also in terms of increased newlinecarbon emissions. As a result, cost reductions associated with energy newlineconservation and effective energy-aware resource management is required for newlinecloud data centres. Dynamic Virtual Machine (VM) consolidation is an newlineeffective method for reducing energy consumption, and it is extensively newlineemployed in large cloud data centers. It achieves energy reductions by newlineconcentrating the workload of active hosts and switching idle hosts into low newlinepower state; moreover, it improves the resource utilization of cloud data newlinecenters. However, the Quality of Service (QoS) guarantee is fundamental for newlinemaintaining dependable services between cloud providers and their customers newlinein the cloud environment. Therefore, reducing the power costs while newlinepreserving the QoS guarantee, and decreasing the number of failures is newlineconsidered as the two main goals of this study. For achieving these three newlinemajor contributions have been performed in this work for cloud data centres newlinewhich are described clearly. newlineFirst contribution of the work, VM consolidation is introduced newlinewhich considers both current and future Uniform Distribution Elephant newlineHerding Optimization (UDEHO) based VM consolidation for resource newlineutilization via host overload detection (Utilization Prediction based Potential newlineOverload Detection (UP-POD)) and host underload detection (Utilization newlinePrediction based Potential Underload Detection (UP-PUD)). UDEHO method newlineefficiently predicts resource use in the future. newline