Review #10: Quantifying the causes of path inflation

From: Rosalia Tungaraza (rltungar@u.washington.edu)
Date: Tue Nov 02 2004 - 23:33:30 PST

  • Next message: Tyler Robison: "Review of "Quantifying the Causes of Path Inflation""

    In this paper, the authors tried to find out the causes of unnecessarily
    long Internet paths. In their investigations they nailed down the problem
    to interactions (or routing mechanisms) within and between Internet
    domains specifically intra-domain, peering points, and inter-domain
    routing. Among these, the Inter-domain routing contributed the most to
    path inflation. They do so due to the policies used by ISPs to facilitate
    routing from one domain to the other. Peering policies are the second
    largest contributor to path inflation. It is argued that in both
    situations ISPs lack good techniques/tools to opt for shorter routing
    paths.

    The authors report that commercial concerns are not the cause of the high
    path inflation resulting from inter-domain routing policies. This is very
    surprising because I expected much of these path inflation problems to be
    a result of decisions made by ISPs based on how much profit they can make
    rather than the underlying network topology. For instance, whenever the
    no-valley and prefer-customer policy is used, paths through customers
    should have the highest inflation followed by paths through peers, while
    paths through providers should have the least. I would also imagine that
    the shortest AS-path to be rarely used unless it is the same as the
    longest path through customers. The paper contradicts my expectations by
    showing that such policies do not contribute to path inflation.

    In general I think the paper was well written. One of the successes of
    this paper is the fact that the authors managed to neatly define/isolate
    the causes of Internet path inflation into six categories and used those
    to analyze the problem.

    A minor improvement to this paper could be achieved by giving more
    information about the tool that the authors believe ISPs need to
    facilitate better peering point and AS-path selection. Hence, future work
    could be geared towards researching how to design and implement such a
    tool and how ISPs could use it to provide shorter paths for their traffic
    while optimizing their profits.


  • Next message: Tyler Robison: "Review of "Quantifying the Causes of Path Inflation""

    This archive was generated by hypermail 2.1.6 : Tue Nov 02 2004 - 23:33:32 PST