面向云计算的能效智能路由理论与技术研究

发布时间：2018-01-07 07:26

本文关键词：面向云计算的能效智能路由理论与技术研究　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着云计算中IT设备数量的不断增加,数据中心的规模也越来越大,云计算环境下的能耗问题日益突出,能耗问题已成为制约互联网和通信业可持续发展的重大障碍。在传统网络系统设计中,主要通过冗余的链路和设备提高网络的可靠性,以应对突发的故障失效,没有考虑到网络能效因素。因此,随着网络规模的扩大和网络设备的不断更新,能耗高、效率低,浪费多等诸多问题是实现节能面临的一个挑战。云计算中各个虚拟机对资源的调度具有随机性、突发性,相对于传统的互联网而言,云计算的网络业务流量具有更大的波动性和时变性。在如此高度动态变化的网络业务下,如何降低网络能耗并提高通信效率是云计算网络面对的主要问题。本文研究面向云计算的高能效智能路由技术,旨在改善网络性能并提高网络能效。在传统骨干网中,IP流的路径的选择主要是基于各种路由协议,这些协议算法实际上是选择资源最大的链路带宽或最小跳数进行通信,这样的选择方式往往与节能的目的背道而驰,所以,仅仅依靠当前的路由协议无法实现云计算的节能问题。本文在云计算网络柔性分离的基础上,提出了三个能效智能路由算法,从路由技术层面开展了面向云计算的能效机制研究。本文主要通过两方面实现能效网络,一是能效路由策略,二是休眠策略。提出了三种基于智能路由技术的路由算法,分别是基于小生境遗传算法的能效路由算法(NGAERA)、基于能效优先和认知理论的路由算法(ECRA)和基于时间序列预测理论和休眠策略的路由算法(ARGMERA)。NGAERA算法通过随机深度优先搜索算法和适应值得到初始种群,然后利用小生境遗传算法进行遗传操作,得到最小网络能耗的全局最优解,但该算法的时间复杂度较高。为此,本文提出ECRA算法,该算法在能效优先思想的基础上,将OSPF路由协议中经典Dijkstra路由算法与认知理论相结合,实现了权重自适应的最小能耗路由算法,使其能够根据网络能耗动态地改变链路权重,保证网络的能耗最小。最后,针对云计算网络业务流量高度动态变化的特性,本文提出了ARGMERA路由算法。该算法分为两个部分,第一部分利用时间序列预测理论建立流量感知模型,第二部分结合ECRA算法建立休眠策略及能效路由策略,实现了具有实用价值的能效路由技术雏形。
[Abstract]:With the increasing number of IT devices in cloud computing, the scale of data center is becoming larger and larger, and the problem of energy consumption in cloud computing environment is becoming more and more prominent. Energy consumption has become a major obstacle to the sustainable development of the Internet and communications industry. In the traditional network system design, the reliability of the network is mainly improved by redundant links and equipment to cope with sudden failure. Therefore, with the expansion of network scale and the continuous updating of network equipment, energy consumption is high and efficiency is low. Many problems, such as waste, are a challenge to realize energy saving. In cloud computing, the scheduling of resources by virtual machines is random, sudden, compared with the traditional Internet. Cloud computing network traffic has greater volatility and time variability. In such a highly dynamic network business. How to reduce network energy consumption and improve communication efficiency is the main problem facing cloud computing network. In order to improve network performance and improve network efficiency, the routing of IP flows in traditional backbone networks is mainly based on various routing protocols. These protocol algorithms actually select the link bandwidth or minimum hops of the largest resource for communication, which often runs counter to the purpose of energy saving, so. Based on the flexible separation of cloud computing network, this paper proposes three energy-efficient intelligent routing algorithms. From the point of view of routing technology, the research of energy efficiency mechanism for cloud computing is carried out. In this paper, energy efficiency network is realized through two aspects, one is energy efficiency routing strategy. The second is dormancy strategy. Three routing algorithms based on intelligent routing technology are proposed, which are energy efficiency routing algorithm based on niche genetic algorithm (NGAERA). Energy efficiency first and Cognitive Theory based routing algorithm (ECRAA) and time Series Prediction Theory and dormancy Strategy based routing algorithm ARGMERAA). The NGAERA algorithm is worth the initial population through random depth-first search algorithm and adaptive algorithm. Then the global optimal solution of minimum network energy consumption is obtained by genetic operation using niche genetic algorithm, but the time complexity of the algorithm is high. Therefore, ECRA algorithm is proposed in this paper. Based on the idea of energy efficiency priority, this algorithm combines the classical Dijkstra routing algorithm in OSPF routing protocol with cognitive theory to realize the weighted adaptive minimum energy consumption routing algorithm. It can dynamically change the link weight according to the network energy consumption to ensure the minimum energy consumption of the network. Finally, aiming at the characteristics of cloud computing network traffic high dynamic change. In this paper, ARGMERA routing algorithm is proposed. The algorithm is divided into two parts. In the first part, the traffic awareness model is established by using the theory of time series prediction. In the second part, based on the ECRA algorithm, the dormancy strategy and the energy efficiency routing strategy are established to realize the prototype of the energy-efficient routing technology with practical value.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP18;TP3

【相似文献】