基于OpenStack的资源动态分配框架的设计与实现

发布时间：2018-05-06 08:34

本文选题：云计算 + 资源优化　；参考：《西北大学》2014年硕士论文

【摘要】：云计算毫无疑问已经成为当前IT业界最为重要且热门的主题,它为传统IT资源的管理和使用方式带来了革命性的创新。随着云用户的日益增加,如何使云数据中心的虚拟化资源在实现容灾的基础上,删除冗余数据,从整个数据中心来讲减少云用户的等待时间,同时还能提高计算资源的利用率,让整个云数据中心的资源得到充分均衡地利用。这已经成为当前最重要也最受关注的云计算环境中资源调度问题之一。本论文主要研究IaaS云平台中的数据中心虚拟化计算资源的动态分配与调度策略,以目前最主流的开源IaaS云平台OpenStack为依托,在前人工作的基础之上展开如下研究工作： (1)分析了当前云计算技术的发展背景和研究意义以及国内外对云计算技术研究的现状,并对云计算的定义、基本特征、服务类型做了详细的介绍。对OpenStack资源调度机制包括资源调度模块、资源调度过程进行了分析研究。 (2)基于OpenStack对IaaS云平台的虚拟化计算资源进行量化。提出并设计了在OpenStack云平台上计算资源的调度算法,包括CPU密集型、内存密集型等调度算法和实时反馈的综合负载均衡资源优化策略。同时阐述了实时反馈的综合负载均衡资源优化策略的主要API的Python定义和保存该策略运行时信息的XML文件的定义。 (3)通过分析开源IaaS云平台OpenStack的资源调度机制,提出一种基于OpenStack的可扩展的多资源池分布式资源调度框架,该框架将物理主机划分成多个逻辑资源池,在全局上针对各个资源池设定全局资源优化策略,在每一个资源池本身设定局部的资源优化策略。实验结果表明,本文提出的多资源池分布式资源调度框架很好地实现了OpenStack的资源优化,并在提高资源利用率的同时实现了性能隔离与安全隔离,具有良好的扩展性。
[Abstract]:Cloud computing has undoubtedly become the most important and hot topic in the current IT industry. It has brought revolutionary innovations for the management and use of traditional IT resources. With the increasing number of cloud users, how to make the virtualized resources of the cloud data center on the basis of disaster recovery and delete redundant data from the whole data center. Reducing the waiting time for cloud users can also improve the utilization of computing resources and make full use of the resources of the whole cloud data center. This has become one of the most important and most concerned cloud computing environment resource scheduling problems.
This paper mainly studies the dynamic allocation and scheduling strategy of the data center virtual computing resources in the IaaS cloud platform. Based on the most mainstream open source IaaS cloud platform OpenStack, the following research work is carried out on the basis of the predecessors' work:
(1) the background and significance of the current cloud computing technology and the status of the research on cloud computing technology at home and abroad are analyzed, and the definition, basic features and service types of cloud computing are introduced in detail. The OpenStack resource scheduling mechanism, including the resource scheduling module and the resource scheduling process, is analyzed and studied.
(2) quantifying the virtual computing resources of IaaS cloud platform based on OpenStack. A scheduling algorithm for computing resources on the OpenStack cloud platform is proposed and designed, including CPU intensive, memory intensive scheduling algorithms and real-time feedback integrated load balancing resource optimization strategies. Meanwhile, the real-time feedback integrated load balancing resource is described. The Python definition of the main API of the optimization strategy and the definition of the XML file for saving the runtime information of the policy.
(3) by analyzing the resource scheduling mechanism of the open source IaaS cloud platform OpenStack, a scalable multi resource pool distributed resource scheduling framework based on OpenStack is proposed. The framework divides the physical host into multiple logical resource pools, setting the whole Bureau resource optimization strategy for each resource pool on the global scale, and setting up each resource pool itself. The experimental results show that the multi resource pool distributed resource scheduling framework proposed in this paper has well implemented the resource optimization of OpenStack, and achieves performance isolation and security isolation at the same time of improving resource utilization, and has good scalability.

【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.01

【参考文献】