云计算中虚拟机节能调度机制研究

发布时间：2018-08-09 16:48

【摘要】：云计算的出现给当代信息产业带来一场变革，通过按需付费的服务模式，云计算能够提供灵活的按需供给的计算资源服务。随着云计算的发展，世界各地都建立起了包含数千计算节点的大型云数据中心。然而，数据中心需要消耗大量的能源，这不仅增加了云服务提供商的运营成本，而且产生了大量的二氧化碳排放，对环境造成了污染。目前，数据中心利用虚拟机动态迁移技术，对用户资源进行整合，能够有效提高资源利用率、降低能耗。但同时，虚拟机在整合过程中会发生服务器主机过载的现象，导致服务性能降低，影响用户的服务质量，甚至会给云服务提供商带来巨大的经济损失。针对上述问题，本研究提出了一种云环境中虚拟机节能调度启发式算法，该算法综合考虑了能量消耗和用户服务质量两方面因素，通过虚拟机整合，提高计算资源的利用率，从而降低数据中心的能耗状况，减少碳排放。同时该算法将户服务等级协议(Service Level Agreement, SLA)的违反率保持在较低的水平，，从而保证了用户的服务质量。本文主要研究工作和成果如下：首先，本课题中虚拟机节能调度算法分为四个部分：主机过载检测、主机低负载检测、虚拟机选择以及主机选择。在主机过载检测的步骤中，通过分析主机过载时，虚拟机动态迁移引起的能耗变化状况，得出一种虚拟机迁移的启发式策略。其次，运用前面得出的启发式策略提出了一种虚拟机节能调度算法，该算法对虚拟机的负载进行预测，从而对主机的过载情况进行判定，进而实行虚拟机的节能调度。最后，通过采集虚拟机负载的真实数据，在CloudSim云模拟平台中对本课题算法进行了对比实验分析。实验结果表明，该算法能够有效的降低数据中心能耗，同时将SLA的违反率保持在降低的水平，取得了较好的效果。
[Abstract]:The emergence of cloud computing has brought a revolution to the modern information industry. Through the on-demand service model, cloud computing can provide flexible on-demand computing resources services. With the development of cloud computing, large cloud data centers with thousands of computing nodes have been established all over the world. However, data centers consume a lot of energy, which not only increases the operating costs of cloud service providers, but also produces a lot of carbon dioxide emissions, which pollutes the environment. At present, using virtual machine dynamic migration technology to integrate user resources, data center can effectively improve resource utilization and reduce energy consumption. But at the same time, the virtual machine in the process of integration will occur server host overload phenomenon, resulting in service performance degradation, affect the quality of service of users, and even bring huge economic losses to cloud service providers. In order to solve the above problems, a heuristic algorithm of virtual machine energy saving scheduling in cloud environment is proposed in this paper. This algorithm considers both energy consumption and user service quality, and improves the utilization ratio of computing resources through virtual machine integration. Thus reducing the energy consumption of the data center, reducing carbon emissions. At the same time, the algorithm keeps the violation rate of (Service Level Agreement, SLA) at a low level, thus ensuring the quality of service of the user. The main research work and results are as follows: firstly, the virtual machine energy-saving scheduling algorithm is divided into four parts: host overload detection, host low load detection, virtual machine selection and host selection. In the process of host overload detection, a heuristic strategy of virtual machine migration is proposed by analyzing the energy consumption change caused by virtual machine dynamic migration when the host is overloaded. Secondly, using the heuristic strategy, a virtual machine energy-saving scheduling algorithm is proposed, which predicts the load of the virtual machine, and then determines the overload of the host, and then implements the energy-saving scheduling of the virtual machine. Finally, by collecting the real data of virtual machine load, the algorithm is compared and analyzed in CloudSim cloud simulation platform. The experimental results show that the proposed algorithm can effectively reduce the energy consumption of the data center, while keeping the SLA violation rate at the reduced level, and achieves good results.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP302

【参考文献】