云计算环境中应用布局与协同关键问题研究

发布时间：2018-06-03 16:14

本文选题：应用布局 + 应用协同　；参考：《山东大学》2015年博士论文

【摘要】：随着互联网的快速发展和普及,出现了越来越多的大型复杂的网络应用,并且逐渐向着网络化、协同化的趋势发展,应用之间的协同需求越来越普遍,如何满足应用之间灵活、高效的协同需求成为目前PaaS研究中的一个热点问题。云计算技术的迅速发展为解决这一问题提供了契机,云计算技术能够为应用提供高可用性,弹性可扩展的能力,然而,如何基于云计算技术灵活、快速、自动的构建高可用的协同应用仍然是当前的一大挑战。作为云计算技术的重要组成部分,PaaS中应用的布局及应用之间的协同问题,已经成为当前的研究热点之一。本文致力于在应用布局过程中提高应用的服务质量、降低云平台的运营成本,自动构建协同元模型。但是,由于云应用高度动态、自治、负载变化快的特点,导致仍然存在以下问题亟待解决：(1)应用的布局与调整。应用在云平台中的布局对云平台的运营成本,应用及协同的执行效率都有着极大的影响,如何使应用的布局能够在满足各自的运行效率要求的前提下,降低应用之间协同成本的同时又要尽量减少云资源的占用,以降低云服务提供商的成本是应用布局和调整时的一个挑战。(2)协同元模型的自动构建。如何利用电子商务云平台中不同ISV所部署不同应用的应用描述,自动、快速的构建能够满足不同用户需求的协同元模型,是当前面临的一大挑战。(3)协同元模型的自动演化。如何根据协同应用的实例化信息,高效的对云平台中的协同元模型进行自动演化,使其更加接近用户的需求,直接关系到协同应用实例化的效率以及用户使用协同应用的复杂度,是协同应用执行过程中的核心问题之一。本文针对云平台中应用协同存在的关键问题展开研究,主要贡献体现在如下几点：(1)针对提高应用性能与降低服务成本之间的矛盾问题,提出一种基于性能-成本均衡的应用布局策略,实现用尽可能低的成本为应用提供更好的服务质量这两个矛盾目标之间的平衡。为了使云服务提供商在应用布局时既能满足各应用的运行效率要求,又能降低应用之间的协同成本且尽量减少所占用的云节点,以降低云服务提供商的成本,需要研究云应用的布局问题。针对这一问题,本文将应用的布局问题建模为一个应用布局图,利用帕雷托最优思想适合于解决多目标矛盾性问题的特点,提出一种基于性能-成本均衡的应用布局策略,有效的实现了应用性能与成本这两个矛盾目标之间的平衡。实验结果表明,该方法能够较好的解决应用布局中性能与成本难以兼顾的问题,找到一个较优的布局方案。(2)针对云平台中应用的动态变化,提出一种应用布局调整方法,在提高应用协同性能的同时,提高了应用布局调整的效率。为了在云平台中应用状态动态变化时能够及时做出调整,保证整个平台中所有应用的运行效率,降低平台中应用的协同成本和运营成本,需要研究云应用布局的调整问题。针对该问题,本文提出一种协同应用布局调整方法,该方法将应用及其协同成本建模为图,根据现有的应用布局方案及各节点负载情况和应用之间的协同成本,在应用运行状态发生变化时,能够快速找到应用布局调整方案。实验结果表明,该方法具有较好的性能,且可伸缩性较强,能够有效的提升应用协同性能。(3) 针对如何从不同ISV提供的大量的应用描述中自动构建协同元模型问题,提出一种基于扩展图规划的协同元模型构建方法,能够在协同元模型构建过程中减小搜索空间,快速的找到符合用户需求的多个协同元模型,提高了协同元模型的构建效率。为了使平台中的应用能够协同进行工作,主动为用户提供协同应用的多个选择,需要研究应用协同元模型的自动构建问题。针对该问题,本文提出一种基于扩展图规划的协同元模型构建方法,该方法通过改进传统的图规划算法,在其中加入辅助节点,减小了搜索空间,并且能够避免多次反向搜索,从而提高协同元模型构建效率。实验结果表明,该方法能够高效、自动的构建协同元模型,并且具有较强的伸缩性,能够自动构建出多个符合不同用户需求的协同元模型,并保证协同元模型构建的正确率。(4)针对协同元模型自动演化问题,提出一种协同元模型的自动演化方法,使演化后的协同元模型更加接近用户需求,提高了协同应用的实例化效率,同时降低了用户使用协同应用的复杂度。为了提高协同应用的实例化效率,同时降低用户使用的复杂度,需要研究协同元模型的自动演化问题。针对该问题,本文提出一种协同元模型的自动演化方法,该方法将协同应用建模为多维空间中的点,利用聚类算法,将协同元模型自动演化为多个符合不同需求的协同元模型,使协同元模型更加接近实际需求。实验结果表明,该方法具有较好的性能,能够显著减少协同应用与协同元模型之间的距离,降低用户使用复杂度与实例化时间。
[Abstract]:With the rapid development and popularization of the Internet, more and more large and complex network applications have emerged, and the trend of networking and synergy is gradually developing. The collaborative demand between applications is becoming more and more common. How to meet the flexible and efficient collaborative requirements between applications has become a hot issue in the current PaaS research. The rapid development of the technology provides an opportunity to solve this problem. Cloud computing technology can provide high availability and resilient ability for applications. However, how to build high available collaborative applications based on cloud computing technology is still a major challenge. As an important component of cloud computing, PaaS is an important component of cloud computing technology. The collaboration problem between application layout and application has become one of the hotspots of current research. This paper aims to improve the service quality of application, reduce the operating cost of the cloud platform, and automatically build a collaborative meta model. However, because of the high dynamic, autonomous and fast changing features of the cloud application, it still exists. The following problems need to be solved: (1) the layout and adjustment of the application. The layout of the application in the cloud platform has a great influence on the operation cost, application and efficiency of the cloud platform. How to make the layout of the application to meet the requirements of their respective operating efficiency and reduce the cooperative cost between applications and as much as possible Reducing the occupancy of cloud resources to reduce the cost of low cloud service providers is a challenge in application layout and adjustment. (2) automatic construction of the collaborative meta model. How to make use of different application descriptions of different applications deployed by different ISV in the e-commerce cloud platform, automatically and quickly build a collaborative meta model that meets the needs of different users. A big challenge before. (3) automatic evolution of synergetic model. How to automatically evolve the cooperative meta model in cloud platform according to the instantiation information of cooperative application, and make it closer to the user's needs, directly related to the efficiency of collaborative application instantiation and the complexity of the user's use of cooperative application. One of the key problems in the execution process is one of the key problems in the application of cloud platform. The main contributions of this paper are as follows: (1) an application layout strategy based on performance cost equilibrium is proposed in order to improve the contradiction between application performance and service cost reduction, which can be used as low as possible. In order to make the cloud service providers not only meet the operating efficiency requirements of various applications, but also reduce the cooperative cost between applications and reduce the cloud nodes occupied by the cloud service providers in order to reduce the cost of the cloud service providers, the cloud service providers need to study the cloud needs. In order to solve this problem, the layout problem of the application is modeled as an application layout graph, and the Pareto optimal idea is used to solve the characteristics of the multi-objective contradiction problem. An application layout strategy based on performance cost equilibrium is proposed, which effectively realms the two contradictory objectives of the application performance and cost. The experimental results show that the method can solve the problem that the application layout neutral energy and the cost are difficult to take into account, and find a better layout scheme. (2) in view of the dynamic changes in the application of the cloud platform, an application layout adjustment method is proposed to improve the application coordination performance and improve the efficiency of the application layout adjustment. In order to apply state dynamic changes in the cloud platform to make adjustments in time, ensure the efficiency of all applications in the whole platform, reduce the cooperative cost and operation cost of the application in the platform, we need to study the problem of the adjustment of the layout of the cloud application. The application and its collaborative cost modeling are illustrated. According to the existing application layout scheme, the load situation of each node and the cooperative cost between the applications, the application layout adjustment scheme can be quickly found when the application state changes. The experimental results show that the method has good performance, strong scalability and can be effectively improved. Application of collaborative performance. (3) in view of how to automatically build cooperative metamodel from a large number of application descriptions provided by different ISV, a cooperative metamodel construction method based on extended graph planning is proposed. It can reduce search space in the construction of collaborative meta model and quickly find multiple collaborative meta models that meet the needs of users. To improve the efficiency of collaborative meta model, in order to make the application of the platform work together, and to provide users with multiple choices of cooperative applications, it is necessary to study the automatic construction of the application cooperative meta model. In this paper, a method of building a collaborative meta model based on extended graph planning is proposed. In the traditional graph planning algorithm, the auxiliary node is added to the algorithm, which reduces the search space, and can avoid multiple reverse search, thus improving the efficiency of the collaborative meta model construction. The experimental results show that the method can efficiently and automatically construct the cooperative meta model, and has strong scalability, and can automatically build multiple conforms to the difference. The cooperative meta model of user requirement ensures the correctness of the cooperative metamodel construction. (4) an automatic evolution method of CO meta model is proposed for the automatic evolution of synergetic meta model, which makes the evolutional synergetic model closer to the user requirement, improves the instantiation efficiency of collaborative application and reduces the user's use of synergy. In order to improve the efficiency of the instantiation of cooperative applications and reduce the complexity of the users, it is necessary to study the automatic evolution of the cooperative metamodel. In this paper, an automatic evolution method of the cooperative meta model is proposed in this paper. The cooperative application is modeled as a point in the multidimensional space, and the clustering algorithm is used to coordinate the collaboration. The metamodel automatically evolves into multiple synergetic models that conforms to different requirements, making the synergetic meta model closer to the actual requirements. The experimental results show that the method has good performance, and can significantly reduce the distance between cooperative application and synergetic meta model, and reduce the users' use of complexity and instantiation time.
【学位授予单位】：山东大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TP393.09

【参考文献】