On Jan 20, 2010, at 2:28 PM, Lars Eggert wrote:
Hi,
Michael asked on the ICCRG list for people to comment whether
MulTFRC should be a WG item here.
My personal (all hats off) belief is that DCCP should focus on work
items that increase DCCP and TFRC deployment. Frankly, I don't think
that "porting" an old and unused idea (MulTCP) to TFRC is going to
lead to that.
I completely agree about your first sentence (but maybe
restricting it to DCCP, not TFRC). And what would increase
deployment? Something that makes the protocol attractive to
its potential users ( = developers of multimedia applications).
The service provided by existing CCIDs such as TFRC ("smooth
rate and TCP-friendly") doesn't seem to be very attractive
from that perspective to me - and MulTFRC adds a lot to that.
I (quite obviously :-) ) disagree about your second sentence.
1) MulTFRC is really much more than merely "porting"
MulTCP to TFRC - one reason why MulTCP might not have
been used is that it simply doesn't work very well, its
operational range is extremely limited (as opposed to
MulTFRC, as we've shown in our CCR paper). Another one
might have been lack of a proper implementation and
specification, and we're trying to address both in case
of MulTFRC.
2) Why isn't it worth trying, especially when we don't
seem to have other proposals on the table that would
foster DCCP deployment?
I'm not opposed to the WG taking this on is they absolutely want. I
mainly don't see a reason for this to be a DCCP WG Experimental RFC
vs. an ICCRG Experimental RFC.
Just to understand, does this "vs." mean that it could
just as well be an ICCRG Experimental RFC in your
opinion, i.e. you're only stating your opinion about these
two choices here, and not about having MulTFRC as an
Experimental RFC at all or not?
Cheers,
Michael
