数据中心网络节能路由研究

发布时间：2018-01-15 04:07

本文关键词：数据中心网络节能路由研究　出处：《清华大学》2013年博士论文　论文类型：学位论文

【摘要】：为了满足云计算应用对网络带宽和可靠性的需求，近年来已提出许多新型的数据中心网络体系结构，它们使用丰富的网络资源互连服务器，以提供高性能和高容错的路由转发服务。然而这些“富连接”体系结构存在着网络能耗巨大和网络能量使用低效等问题，导致了高额的网络营运成本和大量能耗浪费。本文分析典型数据中心体系结构的网络能效，从网络路由和流调度的角度出发研究节省数据中心网络能耗和提高能量使用效率的方法。本文的主要研究内容和贡献如下： (1)网络节能技术研究综述。提出网络节能技术的分类和研究方法，针对典型节能方案的问题模型、实现细节和局限性等进行分析与评价。 (2)数据中心网络节能路由模型研究。分析模块化网络设备的能耗特性，给出网络设备能耗计算模型，为量化评估节能机制的有效性提供支持。提出数据中心网络节能路由模型，，指导节能路由和流调度机制的设计。 (3)数据中心网络能效分析。提出不同数据中心网络体系结构的能效比较方法。通过分析典型数据中心体系结构的网络能耗特点、使用节能机制的效果以及网络性能与能耗间的权衡关系，给出各体系结构网络能效的比较结果。该研究结果能够为设计和升级数据中心网络提供重要的参考依据。 (4)保证网络吞吐率的节能路由机制研究。给出研究问题模型，并证明该问题求解复杂度为NP难。提出一种有效的节能路由算法，在网络吞吐率和可靠性约束下使用网络拓扑剪枝方式计算网络流的节能传输路径。实验结果表明该机制能够有效减少数据中心网络能耗，并且在网络低负载时的节能效果更优。 (5)网络路由与流调度相结合的节能机制研究。研究不同网络流调度和路由策略对数据中心网络能耗的影响，提出一种网络流抢占调度与路由相结合的节能机制，使网络流在其传输路径上独占网络链路带宽，以提高链路利用率和网络能量的使用效率。该机制能够有效节省网络能耗，并有助于改善网络流的平均完成时间。
[Abstract]:In order to meet the demand of network bandwidth and reliability for cloud computing applications, many new data center network architectures have been proposed in recent years, which use abundant network resources interconnection servers. In order to provide high performance and high fault tolerant routing and forwarding services, however, these "rich connection" architectures have many problems such as huge network energy consumption and inefficient network energy use. This paper analyzes the network energy efficiency of typical data center architecture. From the point of view of network routing and flow scheduling, the methods of saving energy consumption and improving energy efficiency of data center network are studied. The main contents and contributions of this paper are as follows: This paper presents the classification and research methods of network energy saving technology, and analyzes and evaluates the problem models, implementation details and limitations of typical energy saving schemes. Secondly, the energy saving routing model of data center network is studied. The energy consumption characteristic of modularized network equipment is analyzed, and the energy consumption calculation model of network equipment is given. In order to quantitatively evaluate the effectiveness of energy-saving mechanism, a data center network energy-saving routing model is proposed to guide the design of energy-saving routing and flow scheduling mechanism. 3) Energy efficiency analysis of data center network. The energy efficiency comparison method of different data center network architecture is put forward. The characteristics of network energy consumption of typical data center network structure are analyzed. The effect of using energy-saving mechanism and the tradeoff between network performance and energy consumption. The comparison results of network energy efficiency of various architectures are given, which can provide an important reference for the design and upgrading of data center network. 4) the research of energy efficient routing mechanism to ensure the throughput of the network. The research model is given, and the complexity of solving the problem is proved to be NP-hard. An effective energy-saving routing algorithm is proposed. Under the constraints of throughput and reliability, the network topology pruning method is used to calculate the energy saving transmission path of the network flow. The experimental results show that the proposed mechanism can effectively reduce the energy consumption of the data center network. And the energy saving effect is better when the network load is low. 5) the energy saving mechanism of network routing and flow scheduling is studied, and the influence of different network flow scheduling and routing strategies on data center network energy consumption is studied. An energy saving mechanism combining preemption scheduling and routing is proposed to enable network flows to monopolize the network link bandwidth in their transmission paths. In order to improve the efficiency of link utilization and network energy, this mechanism can effectively save network energy consumption and help to improve the average completion time of network flow.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TP308;TP393.06

【参考文献】