A Delay-Based Analysis of Multiple Bottleneck Links of End-to-End Paths in the Internet

摘要: Measurement and analysis of bottleneck links play an important role in improving the network quality service (QoS) preventing attacks Internet. Existing methods usually treat link with smallest available bandwidth or largest delay as link, without considering multiple end-to-end path. In this paper, we propose a new approach to measure analyze based on path delay. We design parallel active measurement framework delays many destinations simultaneously. Then algorithm identify is proposed using Ward data clustering method. Experiments are conducted test by measuring 10 different Internet for 14 days. Using approach, have found that mainly few constant which intermediate paths near destinations. Furthermore, results shown number intra-domain takes large portion most cases, hints performance may be greatly influenced iBGP routing. Besides, also demonstrated intercontinental anonymous system (AS) incline paths.