云计算环境中利润最大化的最优服务定价与多服务器系统配置研究

发布时间：2018-06-08 05:15

本文选题：云计算 + 用户感知价值　；参考：《华东师范大学》2017年硕士论文

【摘要】：作为提供计算资源和服务的一种高效方式,云计算越来越受欢迎。由于云计算是一种商业计算模型,因此了解其经济性至关重要。从云服务运营商的角度来看,获取高额利润是最主要的目标,主要取决于给定市场需求下的服务费用和业务成本。为了最大化利润,运营商应该了解其服务费用和业务成本的构成,以及它们是如何由用户的需求分布、应用程序的特性及多服务器系统的配置所决定。本文研究了云计算环境中利润最大化的最优云服务定价和最优多服务器配置问题。在研究过程中,我们对利润最大化问题进行了如下建模。首先,基于用户的感知价值,构建了一个能根据服务价格的变化来反映用户需求分布且符合市场环境中用户需求定律的定价模型。然后,基于用户服务请求对延迟的敏感性,提出了一个MR(Monetary Reward)返现奖励模型。最后,构建了一个基于服务请求响应时间的低服务质量费用补偿模型。为了方便分析,本文将多服务器系统作为一个M/M/m排队模型来处理,并且考虑了两种服务器功耗模式,即idle-speed和constant-speed模式。此外,本文还构建了一个动态自适应闭环控制系统模型(DALCS),使得本文提出的研究方法能动态适应外部条件的改变。研究过程中,本文提出的模型考虑了如下因素:服务请求的数量,服务请求的紧急程度,应用环境的工作负载,多处理器集群的配置,服务等级协议(SLA),服务质量(QoS),用户的服务满意度,低服务质量时的费用补偿,对提交弹性服务请求用户的费用奖励,多服务器系统租赁费用,能耗费用,市场环境中用户的动态需求分布。实验结果表明,本文提出的定价模型符合用户的需求定律。利润最大化的最优服务定价和最优的服务器工作数量及运行速度都能很好的获得。与已有的研究方案OMCPM及UPMR相比,本文构建的利润最大化模型在不同多服务器系统配置下所获得的利润均高于前两者。
[Abstract]:Cloud computing is becoming more and more popular as an efficient way to provide computing resources and services. Since cloud computing is a business computing model, it is important to understand its economy. From the point of view of cloud service operators, obtaining high profits is the main goal, which mainly depends on the service cost and service cost under given market demand. In order to maximize profits, operators should understand the composition of their service costs and business costs, and how they are determined by the distribution of user requirements, the characteristics of applications, and the configuration of multi-server systems. This paper studies the optimal cloud service pricing and the optimal multi-server configuration in cloud computing environment. In the course of the research, we model the problem of profit maximization as follows. Firstly, based on the perceived value of users, a pricing model is constructed, which can reflect the distribution of user demand according to the change of service price and conform to the law of user demand in market environment. Then, based on the sensitivity of user service requests to delay, a reward model of MRM Monetary Rewards is proposed. Finally, a low quality of service cost compensation model based on service request response time is constructed. In order to facilitate the analysis, the multi-server system is treated as an M / M / M / m queueing model, and two server power consumption modes, idle-speed and constant-speed, are considered. In addition, a dynamic adaptive closed-loop control system model is constructed, which makes the proposed research method adapt to the change of external conditions dynamically. In the research process, the following factors are taken into account in the model: the number of service requests, the urgency of service requests, the workload of the application environment, the configuration of multi-processor clusters. Service level Agreement (SLAA), quality of Service (QoS), customer satisfaction, cost compensation for low service quality, cost incentive for submitting flexible service request user, multi-server system rental cost, energy consumption cost, etc. Dynamic demand distribution of users in market environment. The experimental results show that the proposed pricing model conforms to the user's law of demand. The optimal service pricing, the optimal server work quantity and the running speed can be obtained very well. Compared with the existing OMCPM and UPMR schemes, the profit maximization model constructed in this paper is more profitable than the former two models in different multi-server configurations.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【相似文献】