NeTS:Medium:Internet Measurement in the Large

点击翻译按钮获取中文摘要

Despite years of effort, the underlying structure of the Internet is still largely opaque to networking researchers. This opaque structure is also a barrier to ISP administrators trying to diagnose connectivity problems - when a problem occurs, it can be difficult to identify the root cause if it is not in the ISP's own network. In our view, the opaque structure is primarily due to the lack of vantage points from which to measure the Internet: there are more than 100,000 uniquely routed prefixes, yet there are only a few hundred locations from which one can launch probes. This project addresses this lack ofvantage points by developing techniques, using special types of measurement packets, to simulate having a very much larger number of vantage points. Specifically, the PI proposes to build these techniques into a measurement tool, work with several ISPs to have them adopt and deploy the tool, and also deploy it on an NSF and ISP supported testbed called PlanetLab. Using the tool, a number of fundamental questions about the Internet will be studied: How is the Internet evolving? Is the Internet structure self-similar? What is the prevalence of asymmetric routing? Answers to these questions are needed to understand the impact of proposed changes to the Internet. The PI will also make the measurements publicly available. This award is funded under the American Recovery and Reconstruction Act of 2009 (Public Law 111-5).