> But it does mean there's room for someone who wants to understand
> whether an evil burst is 8 packets (as it was in the old days) or
> 8,000.

Ethan Blanton analyzed naturally occurring TCP bursts from three
different networks.  Our results are reported here:

    Ethan Blanton, Mark Allman.  On the Impact of Bursting on TCP
    Performance.  Proceedings of the Workshop for Passive and Active
    Measurement, March 2005.
    http://www.icir.org/mallman/papers/burst-pam05.ps

The results do show clear breakpoints.

Also, Constantinos Dovrolis and his students assess burstiness with
measurement in one or two recent IMC papers (that I don't have handy
reference to right now).

These offer *some* of what Craig is sketching.  As with many things in
networks it is damn near impossible to instrument the real world in such
a way that the full impact of these things is measurable.  So, we're
left with an incomplete story, of course.  But, it's better than no
story.

allman



-- 
Mark Allman -- ICIR/ICSI -- http://www.icir.org/mallman/