|
|
||
George Varghese, Cristian Estan HotNets-II, November 2003 Useful measurement data is badly needed to help monitor and control large networks. Current approaches to solving measurement problems often assume minimal support from routers and protocols (e.g., active measurement) or place the entire burden on the router to support heavyweight mechanisms (e.g., NetFlow). In this paper we argue that the research agenda in measurement must change to consider measurement solutions which enlist the cooperation of routers, protocols, and tools. We believe that the need is so urgent that the deployment issues associated with such holistic solutions can be finessed by cooperation between a few key ISPs and a few key router vendors. If this agenda is accepted, there is a rich vein of technical problems, hitherto considered only from an active measurement perspective, for which there can be new and effective orchestrated solutions. We illustrate this thesis using two examples. Our major example is that of measuring traffic matrices using class counters and class sampling (as opposed to the implementation cost of per-prefix counters, or the errors inherent in tomography). We also provide a smaller example of measuring route stability by modifying route computation. Beyond specific techniques, we hope the guidelines in this paper can provide a focus for discussion among researchers, router vendors, protocol designers, and network operators -- the stakeholders in the measurement enterprise. Paper in PDF and Postscript. |
