Re: [tcpm] [tsvwg] inband signaling (was: Re: Agenda requests for TSVWG@IETF101)

[Toerless Eckert <[e-mail]>]

Thanks, Spencer!

Is there any newer formalistic reference for "TCP friendly CC" or "Internet fair CC"
that one could reference ? Or are we down to the battle ground of coming
up with a random CC algorithm and showing through testing that it will
be within +- 10x speed of some "internet CC mix - Reno, QBR? BBR, nada, ..." ?

If so, is there any defined "Internet CC mix" to do testing with ?

how else... ;-))

Cheers
    Toerless

On Fri, Mar 16, 2018 at 05:54:30PM +0000, Spencer Dawkins at IETF wrote:
> I'm watching this thread, in all its variations across mailing lists, with
> great interest, but not participating, except to say ...
> 
> On Fri, Mar 16, 2018 at 4:44 PM, Toerless Eckert <[e-mail]> wrote:
> 
> > Michael:
> >
> 
> deleted down to
> 
> 
> > Maybe using a rather older refeence CC scheme like reno to define an easily
> > implementable version of the CIR..PIR goal is not the right approach,
> > but the description shuold be more abstract ? Or just use as Emmanual did
> > rely on TFRC as a reference ? Given how i think ( i may be wrong) that
> > TFRC is
> > not really applied to TCP, i thought another approch would have to be used.
> 
> 
> I wouldn't use TFRC as a reference unless you can find people who are
> actively using it as if it was a good thing.
> 
> I was still on the IAB when RMCAT was chartered, but I was also on the IAB
> program committee for https://www.iab.org/activities/workshops/cc-workshop/,
> which was the immediate predecessor activity for RMCAT, and IIRC, we
> couldn't find anyone who actually used TFRC for serious work. It was
> apparently being used as a fig leaf because prior to TFRC we had no advice
> for non-TCP rate control at all. So, at least people could say "do
> something, and TFRC is something", in their drafts without being flushed
> out of the IESG.
> 
> IMO, of course.
> 
> Spencer, speaking NOT as an AD, in this case ...

-- 
---
[e-mail]