[Iccrg] Re: I-D ACTION:draft-irtf-iccrg-welzl-congestion-control-open-research-01.tx t

[Bob Briscoe]

Lachlan,

You're (of course) strictly correct - that I'm a proponent of a 
policing function in the network to regulate e2e congestion control.

But the goal of re-ECN was "What's the minimum necessary in the 
network that's still sufficient to give accountability for causing 
congestion." Only in that sense does it comply with the e2e 
principle. After all, the e2e principle has been clarified as a 
guideline to make designers think long and hard hard before putting 
something in the network.

There is a parallel here with RED and drop-based TCP. Drop is the 
minimum function necessary in the network to make TCP's e2e 
congestion control work. And RED-based drop makes it work 'better' 
(allegedly removing synchronisation etc). So the e2e-ness of TCP is a 
point on a spectrum that isn't quite extreme e2e. And ECN-based TCP 
is a little further inward on the spectrum. And re-ECN is a little 
further in again in order to add accountability.

At least we removed the need for per-flow policing in the network, so 
new congestion controls can evolve - that's the important bit. There 
is still optional per-flow processing in the network, but it can work 
without it.


Bob

At 14:07 02/06/2008, Lachlan Andrew wrote:
>2008/5/30 Bob Briscoe <rbriscoe at jungle.bt.co.uk>:
> >
> > At 15:02 27/05/2008, Michael Scharf wrote:
> >>
> > [snip]
> >>
> >> For instance, [1] concludes that "congestion control is one
> >> function that is not well suited to end-to-end implementation."  I
> >> guess that not everyone will agree to this statement.
> >
> > I for one disagree...
>
>Doesn't it depend on how "end-to-end implementation" is defined?  Bob,
>I thought you championed the idea of re-ECN, which is all about the
>need for network involvement in congestion control.  The end systems
>make the decision, but it isn't "end-to-end" in the sense of
>interaction only between end-to-end peers at layer 4.
>
>Lachlan
>
>--
>Lachlan Andrew Dept of Computer Science, Caltech
>1200 E California Blvd, Mail Code 256-80, Pasadena CA 91125, USA
>Ph: +1 (626) 395-8820 Fax: +1 (626) 568-3603
>http://netlab.caltech.edu/lachlan

____________________________________________________________________________
Bob Briscoe, <bob.briscoe at bt.com>      Networks Research Centre, BT Research
B54/77 Adastral Park,Martlesham Heath,Ipswich,IP5 3RE,UK.    +44 1473 645196