ELVMC: A Predictive Energy-Aware Algorithm for Virtual Machine Consolidation in Cloud Computing

摘要

Virtual machine consolidation (VMC) is a technology that aggregates virtual machines distributed on multiple physical machines into a small number of physical machines to improve resource utilization and energy efficiency of data center. However, excessive virtual machine aggregation and migration can also have a significant negative impact on performance. In this paper, an algorithm named ELVMC with multiple resource prediction is proposed for optimal virtual machine consolidation. It applies a modified Best-Fit Decreasing (BFD) algorithm for resource optimization at both overloaded hosts and underloaded hosts with consideration of load balancing. Different from current research, ELVMC aims to obtain an optimal virtual machine (VM) placement during each consolidation process by simultaneously optimizing multiple system performance metrics in terms of energy consumption, VM migrations and QoS guarantees while keeping the load balanced. Simulation results show that ELVMC is superior to the state of the arts, including the traditional BFD and SABFD-HS algorithms as well as recent research VMCUP-M and MUC-MBFD.