[e2e] Is a control theoretic approach sound?

[Xiaoliang (David) Wei]

Hi Saad,

Not sure if the following may help to explain:

> Consider the following analogy: let a cart being pulled by 50 mice.

This is a very interesting example:)  Let me follow this example:

1) if the cart is very small and light, and if every mouse pulls very hard
and brakes very hard to chase the target speed, then the cart goes as what
you described.
2) if the cart is very heavy, and every mouse pulls slightly, then it takes
a long time for the cart to gain the target speed, and the system may
oscillate around the target speed slightly.
3) If the mice are a little bit cleverer: they consider the speed of the
cart. If the speed is far from the target speed, they pull very hard or
brake very hard; if the speed of the cart is close to the target speed, they
pull slightly or brake slightly, or even stop pulling. Then there is a hope
to make the system stays around the target speed? :) Actually, even 50
elephants come to pull the cart, they have to follow the same rule: pull
or brake slightly when the speed is close to the target.

     I guess we can say that FAST algorithm is doing (3), while Vegas is
doing (2) since Vegas has a fixed step size delta_cwnd=+/-1 during
congestion avoidance. Hence Vegas is not so responsive in high speed
environment.
     Even by doing (3), there may be some problem: how to set the strength
according to the gap from target? If the strength is too aggressive, the
speed of the cart may jump from "much lower than target" to "much higher
than target", and the system oscillates widely. Here, the control theories
may come in to help to design the parameters.

     As in (3), the elephants and mice have no difference in the strength of
pulling and braking. We don't need to worry about how many elephants or
how many mice we have.


> Are there measurements about multiple FAST TCP sharing a unique
> bottleneck? If FAST TCP still stable under these conditions? Is it still
> performant as well?
>
     Yes, there were some experiments with 10 FAST flows (the duration
of the flows >1800sec) of different propagation delays, sharing the same
dummynet router, showing that the FAST still stay around the target
speed.:)

> Any feedback will be most appreciated. Is there something I am missing?
>
> Saad Biaz
> Assistant Professor
> Auburn University
>

-David

Xiaoliang (David) Wei             Graduate Student in CS at Caltech
http://www.cs.caltech.edu/~weixl
====================================================