摘要数据中心网络(Data Center Network)是应用于数据中心内的网络，随着云计算时代数据量的大规模持续性增长，从规模、可拓展性、可靠性、实用性、安全性、高健壮性、低配置开销、设备间的高带宽、网络协议的高效性、灵活的拓扑和链路容量控制等多个方面对数据中心网络提出了更高的要求，同时，数据中心网络中数据传输呈现出很高的峰值吞吐量，并且交换数据集中、具有很大的事件突发性，使得很多问题开始显现出来。随着数据中心网络规模的扩大，需要越来越多的服务器和交换机，交换机与服务器交织组成的网络规模也相应的越来越大，高过载率使得这些交换机很容易成为带宽的瓶颈。为了实现更好的负载均衡和容错，需要从全网的角度来考虑这些问题。75481

本文介绍了基于离散事件的、开源的多协议网络仿真软件--Omnet++, 并在Omnet++平台上搭建了使用TCP协议的模拟数据中心网络，从慢启动、拥塞避免、快速重传、快速恢复多个方面来比较TCP Tahoe 、TCP Reno和TCP NewReno几种拥塞控制算法在数据中心网络中的表现，并分析同时连接同一交换机的客户端数量的多少，超时重传时间的长短，交换机缓存大小等多个参数对数据中心网络中TCP协议传输效率的影响，根据数据中心网络高带宽、低延迟的特点，提出改进TCP协议的方案。

毕业论文关键字  数据中心网络  拥塞控制  omnet++  TCP  协议优化

毕业设计说明书外文摘要

Title    The parameter optimization of TCP protocol in   Data Center Network                 

Data Center Network is a kind of network that applied to the center of Data Center,with the massive sustained growth of cloud computing data,cloud computing puts forward higher requirements to data center from large-scale,scalability, reliability,applicability,safety,high robustness,low profile overhead,high bandwidth between devices,the efficiency of network protocol,flexible topology and link capacity control to the Data Center Network。At the same time,Because of the data transmission of Data Center Network showing high peak thrughput, centralized data exchange and with great sudden events,many problems began to emerge。With the expansion of the data center network,we need more servers and switches,thus one switch will be connected to more and more servers,high overload rate makes these switches can easily become the bottleneck of bandwidth。In order to achieve better Data Center Network load balancing and fault tolerance,we need to think this from the point of view of the whole network。

This article describes the network simulation software--Omnet++,which is based on discrete event,open-source and multi-protocol,and we build a simulation Data Center Network on the Omnet++ platform which is using TCP protocol,to compare the performence of TCP Tahoe,TCP Reno and TCP NewReno from slow start,congestion avoidance,fast recovery,analyze the influences of TCP protocol efficiency in Data Center Network from the number of clients which are connected to the same switch,the length of timeout retransmission and the cache size of switch,according to the high-bandwidth,low-latency characteristics of Data Center Network,put forwarding a proposal to improve the TCP protocol。

Keywords  Data Center Network  congestion control  omnet++  TCP  protocol optimization

目   次

1  引言 1

1。1  课题背景 1

1。2  问题描述 4

1。3  研究意义 4

2  仿真软件介绍