[e2e] TCP flow count estimation

[Jon Crowcroft]

In message <F66A04C29AD9034A8205949AD0C901040194D92A at win-msg-02.wingroup.windep
loy.ntdev.microsoft.com>, "Christian Huitema" typed:

 >>> almost all TCP modeling is for elephants
 
 >>Does not have to be. You can do a statistical analysis of the behavior
 >>of a population of mice, and look at the drop rate or ECN mark rate that
 >>is needed to tame the flock. You can also take a population composed of
 >>mice and elephants, and look at the way a common signal (e.g. the ECN
 >>mark rate) affects the flock and the herd.

Christian 

agree - although if it were as simple as flocking behaviour, looking
at large numbers of mice (i gues you're implying they are coveyed on
IP on avian carriers, right, hence the  birdbrained behaviour)
would be easier than herding cats let alone elephants

 cheers

   jon