基于连续双向拍卖的云计算资源分配研究
发布时间：2019-02-28 09:50
【摘要】：云计算是一种新兴商业计算模型，商业化的特性使其不断关注用户日益多样化的需求，目标旨在按需灵活合理分配资源，提升数据中心资源利用率。成熟虚拟化技术的运用使得资源分配与调度呈现与以往并行分布式计算不同的特性，因而对云计算中资源分配与调度的研究可以推动云计算进一步发展。 本文利用经济学中市场机制具有调动个体积极性，激励资源共享，实现合理资源配置的特点，将连续双向拍卖应用于商业化特性的云计算环境中，设计了基于连续双向拍卖的云计算资源分配策略算法。该算法考虑不同云用户对资源具有多样化的需求，分别为云用户、资源提供者设置公平透明的定价策略：云用户代理在每个时间单位内基于剩余可用资源数与平均剩余时间独立决定竞价值，资源提供者根据自身工作负载提出要价请求。云用户与资源提供者分别提交竞价和要价给拍卖商，拍卖商管理当前的竞价列表与要价列表，匹配最终交易。文中同时考虑少数对云任务截止时间紧迫程度特别高的特殊需求用户，允许这类用户依据实际情况选择预定适合的资源提供者并协商付费价格。最后采用服务等级协议（SLA，Service LevelAgreement）正式确立云用户与云服务提供者之间的服务合约，，指定云用户实现不同QoS方面的服务水平，探讨QoS请求的实现情况。 在介绍分析云计算仿真工具CloudSim的体系结构、技术实现与具体使用方法后，扩展重编译CloudSim平台，对本文策略算法进行仿真实验与性能分析，通过和其他相关调度算法进行比较，表明本文提出的S-CDA算法在一定程度上较好地适应用户对服务质量的不同需求，适合开放复杂的云计算环境。
[Abstract]:Cloud computing is a new business computing model. The commercial characteristics of cloud computing make it pay more and more attention to the increasingly diverse needs of users. The goal is to allocate resources flexibly and reasonably according to demand and to improve the utilization of data center resources. The application of mature virtualization technology makes the resource allocation and scheduling different from the previous parallel distributed computing. Therefore, the research on resource allocation and scheduling in cloud computing can promote the further development of cloud computing. In this paper, the market mechanism in economics has the characteristics of mobilizing individual enthusiasm, encouraging resource sharing, realizing reasonable resource allocation, and applying continuous two-way auction to the cloud computing environment with commercial characteristics. A cloud computing resource allocation strategy algorithm based on continuous two-way auction is designed. The algorithm takes into account the diverse needs of different cloud users for resources, which are cloud users, respectively. The resource provider sets a fair and transparent pricing strategy: the cloud user agent determines the competing value independently on the basis of the number of remaining resources and the average remaining time within each unit of time, and the resource provider makes a request for price according to its workload. Cloud users and resource providers submit bids and asking prices to auctioneers, who manage the current bidding list and asking list to match the final deal. At the same time, we consider a small number of users with special needs that have a very high deadline for cloud tasks, and allow these users to choose the right resource providers according to the actual situation and negotiate the price of payment. Finally, service level agreement (SLA,Service LevelAgreement) is adopted to formally establish the service contract between cloud users and cloud service providers. Cloud users are designated to implement different QoS service levels, and the implementation of QoS requests is discussed. After introducing the architecture, implementation and application of cloud computing simulation tool CloudSim, this paper extends the recompile CloudSim platform, carries on the simulation experiment and performance analysis to the strategy algorithm in this paper, and compares it with other related scheduling algorithms. It is shown that the proposed S-CDA algorithm is suitable for the open and complex cloud computing environment to a certain extent, which can adapt to the different demands of the users for the quality of service to a certain extent.
【学位授予单位】：华东交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP3
本文编号：2431709
[Abstract]:Cloud computing is a new business computing model. The commercial characteristics of cloud computing make it pay more and more attention to the increasingly diverse needs of users. The goal is to allocate resources flexibly and reasonably according to demand and to improve the utilization of data center resources. The application of mature virtualization technology makes the resource allocation and scheduling different from the previous parallel distributed computing. Therefore, the research on resource allocation and scheduling in cloud computing can promote the further development of cloud computing. In this paper, the market mechanism in economics has the characteristics of mobilizing individual enthusiasm, encouraging resource sharing, realizing reasonable resource allocation, and applying continuous two-way auction to the cloud computing environment with commercial characteristics. A cloud computing resource allocation strategy algorithm based on continuous two-way auction is designed. The algorithm takes into account the diverse needs of different cloud users for resources, which are cloud users, respectively. The resource provider sets a fair and transparent pricing strategy: the cloud user agent determines the competing value independently on the basis of the number of remaining resources and the average remaining time within each unit of time, and the resource provider makes a request for price according to its workload. Cloud users and resource providers submit bids and asking prices to auctioneers, who manage the current bidding list and asking list to match the final deal. At the same time, we consider a small number of users with special needs that have a very high deadline for cloud tasks, and allow these users to choose the right resource providers according to the actual situation and negotiate the price of payment. Finally, service level agreement (SLA,Service LevelAgreement) is adopted to formally establish the service contract between cloud users and cloud service providers. Cloud users are designated to implement different QoS service levels, and the implementation of QoS requests is discussed. After introducing the architecture, implementation and application of cloud computing simulation tool CloudSim, this paper extends the recompile CloudSim platform, carries on the simulation experiment and performance analysis to the strategy algorithm in this paper, and compares it with other related scheduling algorithms. It is shown that the proposed S-CDA algorithm is suitable for the open and complex cloud computing environment to a certain extent, which can adapt to the different demands of the users for the quality of service to a certain extent.
【学位授予单位】：华东交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP3
【参考文献】
相关期刊论文 前7条
1 陈冬娥;杨扬;刘丽;;基于反拍卖的计算网格资源分配方法[J];北京科技大学学报;2007年05期
2 吴吉义;平玲娣;潘雪增;李卓;;云计算:从概念到平台[J];电信科学;2009年12期
3 华夏渝;郑骏;胡文心;;基于云计算环境的蚁群优化计算资源分配算法[J];华东师范大学学报(自然科学版);2010年01期
4 翁楚良;陆鑫达;;一种基于双向拍卖机制的计算网格资源分配方法[J];计算机学报;2006年06期
5 林闯;;《物联网关键理论与技术》专题 前言[J];计算机学报;2011年05期
6 陈康;郑纬民;;云计算:系统实例与研究现状[J];软件学报;2009年05期
7 黄智维;倪子伟;;网格计算环境下资源管理的研究[J];计算机技术与发展;2009年03期
相关硕士学位论文 前2条
1 杨炼;网格资源三层调度模型与算法研究[D];中南大学;2008年
2 赵春燕;云环境下作业调度算法研究与实现[D];北京交通大学;2009年
本文编号：2431709
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2431709.html
