[Iccrg] Meeting agenda (really! :-) )

[Doan B. Hoang]

Michael Welzl wrote:

>>>So, does everyone agree that future congestion control protocols must
>>>be implicit, TCP friendly, and primarily
>>>delay-based with fail-over to a standard loss-based algorithm?
>>>      
>>>
>>Actually I disagree with all of these conclusions.
>>    
>>
>
>and so will many others, I suppose.
>  
>
Implicit, explicit controls are all relative. If you get explicit 
information for your control algorithm at a minimum cost, you
will certainly go with explicit.
Implicit solution for large networks takes a long tme to converge and is 
not applicable for short flows.
I would not rule out explicit congestion control protocols.

Delay-based algorithm is rather tricky as users and their traffic have 
different time constants. fair to one is always
unfair to another.

>My point was that writing a survey document is useless because
>that has been done before. We could take an existing
>one and try to draw conclusions from it (just like we're
>doing now, by criticizing conclusions of others). For this,
>we could also use a document which doesn't arrive at any
>conclusions at all but just presents a overview:
>http://www.welzl.at/research/publications/GFD.55.pdf
>
>... but repeating this effort seems to be a waste of time to
>me, no matter if we agree with the conclusions made in one
>of these documents or not.
>  
>
As I pointed out in one of the previous emails that everyone is tired of 
all the TCP variants
which are mostly useless or of dubious value and the someone else is 
summarizing all these
variants!!!

I sugest that we think up a list of approaches that have not been tried 
before with a short
discussion on each item on the list. It does not matter how peculiar the 
approaches that we
come up. I believe this is more exciting than another summary or survey, 
yet potentially
opens up new solutions (or no go solutions). For example, the 
"cost-based" approach promoted by Bob Briscoe
is one.

>Cheers,
>Michael
>
>
>_______________________________________________
>Iccrg mailing list
>Iccrg at cs.ucl.ac.uk
>http://oakham.cs.ucl.ac.uk/mailman/listinfo/iccrg
>  
>
Cheers,
Doan

-- 
Doan B. Hoang
URL: http://research.it.uts.edu.au/arn