[e2e] Re: [pilc] Bidirectionnal traffic over assymetric link
Dr G Fairhurst
gorry at erg.abdn.ac.uk
Tue Jul 23 08:41:09 PDT 2002
One thing you may like to look at is the internet draft developed
by the PILC WG on this topic.
The draft is at:
and has been recommended to the IESG for publication as a RFC.
The draft talks of many of the issues you mention (plus a few more).
If you'd like to know more, or have further questions, you are welcome
to email me directly, and I'll try to help.
Fatma Louati wrote:
> I have made some simulations of bidirectionnal TCP traffic passing
> through an assymmetric link, evolving several simultaneous connections
> in both direction. It seems that the combination of using AckFiltering
> and Ack Reconstruction, AckFirst scheduling and an Active Queue
> Management like RED gives the best results in term of global throughput.
> In fact the asymmetry that forward connections encounter necessitate the
> use of AF/AR, however the presence of traffic in the other direction
> implies the use of an AckFirst scheduling to protect ACKs against Data
> packets. But limited buffer size will result in a ACK packets
> monopolization. That's why I thought about using a AQM mechanisms in the
> router. This solution gives also an acceptable fairness among the
> different flows.
> Before I go further in this direction I would like to know what do you
> think about it.
> Could you please tell me if you are aware of recent work on this domain.
> Fatma Louati -
> PhD Student - INRIA Sophia Antipolis
> http://www.inria.fr/planete/flouati <http://www.inria.fr/planete/dabbous>
> ps: Simulations topology is accessible in
> pilc mailing list
> pilc at ietf.org
More information about the end2end-interest