Paper Review #10: Quantifying the Causes of Path Inflation

From: Yuhan Cai (yuhancai@cs.washington.edu)
Date: Tue Nov 02 2004 - 22:45:18 PST

  • Next message: Michelle Liu: "Review of "Quantifying the Causes of Path Inflation""

    Title: Quantifying the Causes of Path Inflation
    Authors: Neil Spring, Ratul Mahajan and Thomas Anderson
    Reviewed by: Yuhan Cai

    Main results of the paper:

    , It employs novel techniques to identify the root causes of internet path inflation.

    , It analyzes observed and hypothetical topologies and policies to exploit the effects of different factors.

    Strengths of the paper:

    , It provides the first set of experimental data on the prevalence of late-exit, micro-engineered, and load-balanced policies.

    , Inference techniques are developed for intra-domain routing and peering policies to evaluate them in terms of path inflation.

    , The effects of the 6 individual causes of path inflation are identified and isolated.

    Key limitations:

    , The research is trace-driven, and therefore the results are limited in the sense that only a certain set of ISPs and vantage points is available.

    , The provisioning and policy choices below the IP layer are not quite visible.

    Relevance of the paper:

    , As no existing work has completely explained the effects of various factors on path inflation, this paper characterizes the extent to which different issues affect inflation and presents a useful insight into network and protocol design.

    Future work:

    , The problem of path inflation could be alleviated if better peering point and AS-path selection mechanisms were developed.

    , It would be necessary to consider the generalization of the current framework to other vantage points.


  • Next message: Michelle Liu: "Review of "Quantifying the Causes of Path Inflation""

    This archive was generated by hypermail 2.1.6 : Tue Nov 02 2004 - 22:45:23 PST