面向网格应用的光网络资源分配技术研究

发布时间：2018-09-13 06:16

【摘要】：网格是一种实现资源共享的基础设施。它通过互联网将地理上分布的、异构的资源连接起来，形成一台超级计算机。随着网格技术的发展，网格对网络的能力也提出了较高的要求，光网络能够满足网格提出的要求，便产生了光子网格技术。光网络作为光子网格资源的一部分，它的大带宽、低时延等特性能够有效地解决网格任务的传输问题，同时大大提高了网格任务的处理速度。传统的作业处理过程都是在一台计算机上完成，但是在网格系统中，一个网格作业的处理可能需要将作业分割成多个子任务，分配给资源站点进行处理。那么，如何更好、更经济的分配光子网格资源，，便成为了一个重要的问题。在许多文献中已经提到了很多不同的策略，用来发现和分配满足用户性能需要的资源，但是很少会考虑经济性的分配原则。本文在光子网格的理论基础之上，主要针对“面向网格应用的光网络资源分配”这一关键问题进行研究。本文介绍了光子网格发展和现状、以及资源分配存在的几个问题。针对资源分配的经济性问题，给出了两种都基于“沃格尔”算法改进的资源分配方法。首先，在只考虑网络资源分配（光链路分配）的情况下给出一种改进的算法。通过与其他算法的比较发现改进算法的总计算代价结果会更小，则网格用户在利用网络资源的时候花费较小。其次，在考虑所有网格资源的情况下给出了一种联合最小代价算法。文章中对这种方法的有效性和经济性进行了分析，并通过与其他算法加以比较，发现这种联合最小代价算法的经济性会更好。结果表明，这两种改进的算法能够实现对资源的合理分配，并使用户获得更低代价的性能体验。
[Abstract]:Grid is an infrastructure for resource sharing. It connects geographically distributed, heterogeneous resources over the Internet to form a supercomputer. With the development of grid technology, the grid has put forward higher requirements for the ability of the network. The optical network can meet the requirements of the grid, so the photonic grid technology is produced. As a part of photonic grid resource, optical network has the characteristics of large bandwidth and low delay, which can effectively solve the transmission problem of grid tasks and greatly improve the processing speed of grid tasks. The traditional process of job processing is done on a single computer, but in a grid system, the processing of a grid job may require that the job be divided into several sub-tasks and assigned to the resource site for processing. Therefore, how to allocate photonic grid resources more economically has become an important problem. Many different strategies have been mentioned in many literatures to find and allocate resources that meet the performance needs of users, but economic allocation principles are rarely considered. Based on the theory of photonic grid, this paper focuses on the key problem of resource allocation for grid applications in optical networks. This paper introduces the development and present situation of photonic grid and some problems in resource allocation. Aiming at the economic problem of resource allocation, two improved resource allocation methods based on "Vogel" algorithm are presented. Firstly, an improved algorithm is proposed under the condition that only network resource allocation (optical link allocation) is considered. Compared with other algorithms, it is found that the total computational cost of the improved algorithm will be smaller, and the grid users will spend less on the utilization of network resources. Secondly, a joint minimum cost algorithm is proposed considering all grid resources. In this paper, the effectiveness and economy of this method are analyzed, and compared with other algorithms, it is found that this joint minimum cost algorithm is more economical. The results show that the two improved algorithms can achieve a reasonable allocation of resources and enable users to obtain a lower cost performance experience.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.07

【参考文献】