基于车联网应用的云平台资源调度问题的研究

发布时间：2018-06-12 23:06

本文选题：车联网 + 云计算　；参考：《大连理工大学》2014年硕士论文

【摘要】：基于车联网应用的云计算支撑平台利用虚拟化技术将不同类型的物理服务器和虚拟机等异构资源整合成一个虚拟资源池,按需为不同的用户提供不同类型的车联网应用服务。车联网大部分具体应用服务具有实时性强等特点,这些特点对车联网云平台提出了更高层次的要求。资源调度一直是云计算研究的重要课题之一,其调度问题的好坏直接影响到云平台的稳定性和服务的可靠性。因此,研究车联网云平台上的资源调度算法有着非常重要的理论和现实意义。本文在车联网应用背景下,对云计算的资源调度问题进行以下两个方面的研究。 (1)车联网应用具有多用户、多业务、高并发等特点。为了保障车联网应用在云平台上快速、稳定和可靠的运行,在云计算的基础上,本文提出一种基于车联网应用的MCT-LB-GSA (Minimum Completion Time-Load Balance-Greedy Scheduling Algorithm)任务调度算法。算法以虚拟机资源的当前负载作为约束条件,依照贪心策略将任务调度到当前负载较轻且具有最小任务完成时间上的虚拟机资源上。在C1oudSim环境下进行了实验仿真,实验结果证明：该算法在保证最优任务调度跨度的同时也有效地实现了资源负载均衡,提高了资源利用率。 (2)车联网应用的特点与云计算数据中心物理主机配置的不一致通常会引起负载不均衡。针对该问题,本文提出一种基于车联网的CDM-CU (Combining the distance matching and comprehensive utilization)虚拟机部署算法。本算法并不单纯追求虚拟机和物理服务器性能向量的最优距离,也不单纯追求数据中心的最小负载,而是通过调和因子将二者灵活融合在一起,为用户提交的业务选择合适的物理主机来部署相应的虚拟机集。在CloudS im环境下进行了实验仿真,实验结果证明：该算法能在满足个性化业务的基础上取得很好的负载均衡。
[Abstract]:The cloud computing support platform based on vehicle networking applications integrates different types of heterogeneous resources such as physical servers and virtual machines into a virtual resource pool using virtualization technology and provides different types of vehicle networking application services to different users as needed. Most of the specific application services of vehicle networking have the characteristics of strong real-time, these characteristics put forward a higher level of requirements for the cloud platform of vehicle networking. Resource scheduling is one of the most important research topics in cloud computing. The quality of resource scheduling directly affects the stability of cloud platform and the reliability of service. Therefore, it is of great theoretical and practical significance to study the resource scheduling algorithm on the vehicle network cloud platform. In this paper, the following two aspects of resource scheduling of cloud computing are studied in the context of vehicle networking application. (1) vehicle networking applications have the characteristics of multi-user, multi-service, high concurrency and so on. In order to ensure the fast, stable and reliable operation of the vehicle networking application on the cloud platform, this paper presents a task scheduling algorithm based on cloud computing for MCT-LB-GSA / Minimum completion Time-Balance-Greedy scheduling algorithm. The algorithm takes the current load of the virtual machine resource as the constraint and schedules the task to the virtual machine resource with the lighter load and the minimum task completion time according to the greedy strategy. The simulation results under C1oudSim environment show that the proposed algorithm not only ensures the optimal task scheduling span but also realizes resource load balancing effectively. The characteristics of the vehicle network application are not consistent with the physical host configuration of the cloud computing data center, which usually lead to load imbalance. To solve this problem, a CDM-CU Combining the distance matching and comprehensive utilization) virtual machine deployment algorithm based on vehicle networking is proposed in this paper. This algorithm does not simply pursue the optimal distance between the virtual machine and the physical server performance vector, nor the minimum load of the data center, but combines the two factors flexibly through the reconciliation factor. Select the appropriate physical host for the business submitted by the user to deploy the corresponding virtual machine set. Experimental results under CloudSim environment show that the proposed algorithm can achieve good load balancing on the basis of satisfying personalized services.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09;TP391.44;TN929.5

【参考文献】