===================================================
From takhoa@rice.edu  Wed Feb 11 19:03:12 2004
===================================================

I find that many of the results in the paper are a little bit ambiguous.
This is largely the result of the authors having to extrapolate and infer
the internal router problems from the end-to-end results.  Some of their
calculation methods, for example, using the mean time between 2 end points
for the measurement time, are not accurate.  For instance, if the two end
points have time of 6:00am and 6:00pm, that would make the measured time
as midnight.  Other calculation problems include deciding which routers
are responsible for pathologies in traceroutes.

However, there are a lot of good conclusions coming out of the paper.  For
example, the paper presents the degree of route oscillation and the
consistency of routes fairly well.

This paper does not offer much insights, or solutions into why the
problems occur, or how to fix them.  However, the paper is a good
framework for those who want to address each individual routing behavior
and find out