期刊
IEEE ACCESS
卷 10, 期 -, 页码 81787-81804出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2022.3196240
关键词
Cloud computing; Energy consumption; Data centers; Heuristic algorithms; Costs; Virtual machining; Energy efficiency; Green data centers; service level agreement; virtual machine placement; virtual machine consolidation
This paper proposes a novel approach to improve physical machine efficiency in data centers. By employing heuristics and meta-heuristic algorithms along with eight performance criteria, the approach optimizes power usage in small to medium scale data centers. The results demonstrate significant improvement in energy consumption, number of SLA violations, and number of VMs migrations compared to previous algorithms.
Increasing demand for computational resource as services over the internet has led to the expansion of datacenter infrastructures. Thus, datacenter authorities are striving to adopt optimal power usage schemes to minimize costs, emissions and Service Level Agreement (SLA) violations in their task scheduling for heterogeneous computation centers. One of the most effective strategies to reduce datacenter energy consumption is to maximize the utilization of physical machines and shut down the idle ones. This can be realized through two main algorithms, namely virtual machine placement and virtual machine consolidation. The VM placement method is a dynamic process to put these virtual devices on physical machines. The consolidation technique, however, tries to improve physical machine efficiency through grouping and live migration of dispersed virtual machines on lower number of active physical machine. In this paper, a novel approach is proposed for improving the physical machine efficiency. The approach employs heuristics and meta-heuristic algorithms with eight performance criteria and is implemented on small to medium scale data centers using simulated cloud module. The results indicates that the proposed method showed up to 10.3%, 5.3%, and 12.5% the more significant efficiency rather best previous algorithms, respectively, in terms of the energy consumption, number of SLA violation and number of VMs migration.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据