中文摘要：

云计算数据中心网络的流量特征是研究和设计云计算网络的基础，现有的流量测量研究方法通常要求交换机支持额外功能模块或具备可编程能力，而目前大多数云计算数据中心网络的交换机并不满足此要求。提出一种基于网络层析技术的端到端流量推理算法，仅使用交换机普遍支持的SNMP（简单的网络管理协议）数据，就能快速准确地计算出端到端的流量信息。并通过仿真实验与已有的网络层析算法进行比较，结果表明新算法更适用于大规模的云计算数据中心网络，可以在较短的时间内得到更准确的计算结果，从而为云计算网络的设计和研究提供了重要的参考依据。

英文摘要：

Traffic characteristic of cloud computing data centre networks is the basis of research and design of cloud computing networks. Current study methods of traffic measurement techniques usually ask the switches supporting additional functional modules or being program- mable, however not many of the cloud computing data centre networks can afford such switches. In this paper, we proposed an end-to-end traffic inference algorithm which is based on network tomography. It can rapidly calculate the end-to-end traffic information with high accuracy by only utilising SNMP （simple network management protocol） data ubiquitously supported by switches. Through simulation experiment the algorithm was compared with existing network tomography algorithm, and result showed that the new algorithm was more applicable to large- scale cloud computing data centre networks, and could gain more accurate computing results in short period, so as to provide important refer- ence basis for the design and research of Cloud computing networks.