云计算资源分配机制设计及博弈分析

发布时间：2018-03-12 19:52

本文选题：云计算　切入点：博弈论　出处：《浙江师范大学》2014年硕士论文　论文类型：学位论文

【摘要】：云计算是一种新兴的计算模式,是在超大规模的分布式环境下利用共享基础架构的方法,向云用户提供计算资源、数据存储和网络服务。云用户(企业或个体)通过使用云计算服务,不仅能够降低开销,而且可以专注于自身的核心业务发展,不用担心拥有的基础设备是否能够支持。随着云计算技术的发展,云计算资源的分配和管理成了关键的问题之一,合理的云资源分配机制不仅能够提高云资源的利用率,而且能够更好地满足不同用户的需求,从而使得云资源服务市场稳定有序。传统的云计算资源分配机制是按照用户的订购需求分配,用户按其使用量支付相应的费用,这也是商家起初宣传的商业模式。随着云计算商业领域的发展以及用户的需求日益呈现多样化,云资源提供商开始尝试着不同的云资源分配机制,如带预留的分配,在线竞拍云资源等,更合理地考虑了用户的个性化需求。本文假设每个云用户都是理性的、智能的,建立了云资源分配相关模型,借助博弈论、机制设计及相关理论,对云计算资源分配机制进行了研究,工作内容和创新点如下： (1)在传统的按需分配云资源的分配机制下,结合着混合云环境,建立了混合云工作量分解博弈模型,从性能类型分析和市场类型分析两个方面对混合云用户分解工作量的策略进行了均衡分析。在性能类型分析中证明了博弈中纳什均衡的存在和唯一性,并根据博弈分析得到博弈中纳什均衡的算法,在市场类型分析中得到纳什均衡策略的表达式,并列举数据对纳什均衡策略进行了分析比较。 (2)针对云用户对云资源的请求具有很强的动态性,区别于传统的云资源分配机制,本文从云提供商的角度设计了一个云资源在线竞拍机制。该设计的在线竞拍机制能够满足用户的个性化需求,同时满足激励兼容的性质,并与传统的离线最优竞拍机制进行了对比、模拟仿真,显示了设计的在线竞拍机制具有较好的性能。目前,云计算正朝着多元化方向发展,在我国对云计算资源分配机制研究还存在缺陷的背景下,本文所做工作对我国云计算服务的发展具有一定的参考价值。
[Abstract]:Cloud computing is a new computing model, which provides computing resources to cloud users by using shared infrastructure in a large scale distributed environment. Data storage and network services. Cloud users (enterprises or individuals) can not only reduce overhead but also focus on their core business development by using cloud computing services. With the development of cloud computing technology, the allocation and management of cloud computing resources has become one of the key problems, reasonable cloud resource allocation mechanism can not only improve the utilization of cloud resources, Moreover, it can better meet the needs of different users, thus making the cloud resource service market stable and orderly. The traditional cloud computing resource allocation mechanism is allocated according to the user's order demand, and the user pays the corresponding fee according to their usage. This is also the business model that merchants initially promoted. With the development of cloud computing business and the increasing diversity of users' needs, cloud resource providers are beginning to try different cloud resource allocation mechanisms, such as allocation with reservation. Based on the assumption that every cloud user is rational and intelligent, a cloud resource allocation model is established, with the help of game theory, mechanism design and related theories. The resource allocation mechanism of cloud computing is studied. The contents and innovations are as follows:. 1) under the traditional allocation mechanism of cloud resources according to demand, combined with mixed cloud environment, a mixed cloud workload decomposition game model is established. In this paper, the strategy of decomposing workload of mixed cloud users is analyzed in terms of performance type analysis and market type analysis, and the existence and uniqueness of Nash equilibrium are proved in the performance type analysis. According to the game analysis, the algorithm of Nash equilibrium in game is obtained, and the expression of Nash equilibrium strategy in market type analysis is obtained, and the Nash equilibrium strategy is analyzed and compared by enumerating the data. (2) aiming at cloud users' request for cloud resources, it is very dynamic, which is different from the traditional cloud resource allocation mechanism. This paper designs an online bidding mechanism for cloud resources from the point of view of cloud provider. Compared with the traditional off-line optimal bidding mechanism, the simulation results show that the designed online bidding mechanism has better performance. At present, cloud computing is developing in the direction of diversification. Under the background of the research on the resource allocation mechanism of cloud computing in China, the work done in this paper has certain reference value for the development of cloud computing services in China.
【学位授予单位】：浙江师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】