Cell Relay Archive

[Laurent Jaussi <laurent.jaussi@jaussi.ch>]

Hi,

In the ACTS project EXPERT, we have made an experimental study on 
the performance of TCP over UBR with EPD (several TCP sources
multiplexed at a bottleneck). It turns out that with a good TCP
implemantation you need 
little more than one TCP window size to achieve good performance
in terms of goodput and fairness.

A report on these experiments can be found at:
http://tcomwww.epfl.ch/~jaussi/icatm98.pdf

The length of the burts that arrives to the buffer naturally depends 
on the IP MTU which limits the size of the IP packets, and also on the
TCP window size and TCP acknowledgement strategy (delayed or non-delayed
ACK).


Cheers,

Laurent

Patrick Hurley wrote:
> 
> Has anyone done research or know of research that exists that studies
> the number of cells that a queue must tolerated before it drops IP
> packets?
> 
> I have seen distributions where it shows that most (average) bursts are
> for 5 cells and the maximum burst is for 93 or 100 cells.
> 
> I know of the self-similar paper, that is not what I am looking for.