Re: [PCN] Termination of flows in small aggregates

[Michael Menth <menth at informatik.uni-wuerzburg.de>]

Hi Rüdiger,

you are right that small overtermination can be problematic with small 
aggregates. This has been studied in Section 4.6 in
http://www3.informatik.uni-wuerzburg.de/~menth/Publications/papers/Menth08-Sub-9.pdf
and solutions that can be locally implemented have been proposed and 
demonstrated.

Regards,

   Michael

Ruediger.Geib at telekom.de schrieb:
> Hi Phil,
>
> I'm not a staunch supporter of termination. I'm having a question related 
> to termination of small aggregates.
>
> Assume 10% excess marked traffic at a particular egress.
>
> - 5 flows for IEA1
> - 3 flows for IEA2
>
> Each flow consumes the same bandwidth. 
>
> I don't think that it matters whether decisions are taken at ingress or 
> egress. The farther congestion is from an ingress or egress, the lower 
> the number of IEAs. That means, that either the ingress or egress may 
> control a number of small aggregate IEAs, depending on the location of 
> the pre-congested interface.
>
> As far as I understand, there's no way to signal which flows are the most 
> burdensome for a pre-congested interface. Maybe the above 8 flows consume
> 10 times the average bandwidth of any other flow passing the pre-congested 
> interface. The egress can't learn about that.
>
> I'm not out for an endemic fairness discussion and I don't want to start 
> a re-ECN discussion here. A pragmatic rule then would be "terminate 
> one flow"? Or should the reaction be configurable?
>
> I'm not looking for solutions which result in flow awareness at 
> pre congested interfaces or centralised systems having an overview.
>
> Regards,
>
> Ruediger
>
>
>
> Deutsche Telekom Netzproduktion GmbH 
> Zentrum Technik Einführung 
> Technik Internet Backbone, TE142-19
> Rüdiger Geib
> Heinrich Hertz Str. 3-7
> 64297 Darmstadt
> Tel.: 06151/6282747
> Fax: 0251/7985109
>
>
> Deutsche Telekom Netzproduktion GmbH 
> Aufsichtsrat: Dr. Steffen Roehn (Vorsitzender) 
> Geschäftsführung: Dr. Bruno Jacobfeuerborn (Vorsitzender), Albert Matheis, Klaus Peren 
> Handelsregister: Amtsgericht Bonn HRB 14190 
> Sitz der Gesellschaft: Bonn 
> USt-IdNr.: DE 814645262
>
>
>
>
>
>
>
>
>
> _______________________________________________
> PCN mailing list
> PCN at ietf.org
> https://www.ietf.org/mailman/listinfo/pcn
>   

--
Dr. Michael Menth, Assistant Professor
University of Wuerzburg, Institute of Computer Science
Am Hubland, D-97074 Wuerzburg, Germany, room B206
phone: (+49)-931/31-86644 (new), fax: (+49)-931/888-6632
mailto:menth at informatik.uni-wuerzburg.de
http://www3.informatik.uni-wuerzburg.de/research/ngn