基于包对原理测量瓶颈带宽方法的实现与分析

IMPLEMENTATION AND ANALYSIS OF BOTTLENECK BANDWIDTH MEASUREMENT METHODS BASED ON PACKET PAIR PRINCIPLE

瓶颈带宽是网络测量中非常重要的性能指标。当前大部分测量瓶颈带宽的方法都是基于数据包对PP(Packet Pair)原理。阐述基于此原理的三种测量瓶颈带宽方法并在Linux环境下对三种方法进行实现。通过在搭建的网络环境中测试得出的一系列数据,证明了包对原理的准确性以及背景流量CT(Cross Traffic)对瓶颈带宽测量的干扰。根据测试结果从准确性、部署难易程度等方面分析和比较了三种方法。最后通过实验进一步分析了探测数据包的长度对测量结果的影响。结论可为不同网络应用场景下选择瓶颈带宽测量方法提供有效的参考。

Bottleneck bandwidth is a very important performance index of network measurement,providing important information for the operation and maintenance of network.Many of the current bottleneck bandwidth measurement methods are based on the packet pair principle.The paper details three methods for measuring the bottleneck bandwidth based on the principle.We also program them on Linux platform.The results obtained from measurement prove the accuracy of the packet pair principle and interference caused by cross traffic in the process.The three methods are compared in terms of accuracy,ease of deployment and other aspects.In addition,the paper investigates the influence of the probing packets length on the measurement by experiments.The conclusions of the paper can provide an effective reference for selecting the bottleneck bandwidth measurement method in different network scenarios.