Re: [re-ECN] Pls respond: Do we have a Congestion Exposure community?

[Bob Briscoe <[e-mail]>]

Rolf,

At 19:03 04/09/2009, Rolf Winter wrote:
>I personally am interested in this. I think re-ECN is an interesting 
>endeavor and has significant potential.

Thanks

>Regarding the experiments you mentioned in an earlier Email, do you 
>have something in particular mind how these might look like?

Nope - that's up to you, of course. Do you have something in mind?

Altho I'm open, here's some thoughts:

1/ I know the Internet2 guys (at least the CEO) want to see if 
deploying re-ECN will help a problem they have. They have users who 
want to transfer large chunks of scientific data. Internet2 prevents 
them using just any transport protocol in case they blow other users 
out of the water. They have set up a competition between 
connectionless and connection-oriented. In the latter case users can 
request a nailed up pipe between A & B for their experiment. In the 
former case they have to negotiate all the blocks and throttles that 
may have been deployed between A & B.

The idea is to see whether a generic congestion policing function 
built on re-ECN would solve this dilemma.

2/ We have folks in BT who we're discussing the possibility of using 
congestion in customer contracts. One idea is to trial it in the 
market - but there's a lot of technology trials to go through first.

3/ I'd like to see trials on novel congestion control algos, using 
congestion policing as a permit to allow them not to be TCP-friendly.

4/ other...


Bob


________________________________________________________________
Bob Briscoe,               Networks Research Centre, BT Research