Hi Pasi, Tom, [and other interested DCCP folks,]
In response to your "DCCP - What now?" presentation in tsvarea today,
I'd like to suggest that we meet and talk further about how DCCP might
fit as a "Flow Layer" in the Internet architecture, atop which other
"Semantic Layer" transports would run, as we proposed in our talk. It
seems like DCCP's current main weakness is that it's just not
essential enough for many common applications - but in my view it
_should_ be essential as a component underlying most or _all_
congestion-controlled transports in a next-generation Internet
architecture, even if applications might then rarely use DCCP directly
(although they presumably still could).
For you and any others interested in further discussing broader issues
of transport architecture evolution such as these:
(1) There's an IETF mailing list we started a while ago for these
types of discussions:
"Transport Architecture Evolution"
https://www.ietf.org/mailman/listinfo/tae
(2) Some of us are getting together for an unofficial "bar-BOF"
tomorrow (Tuesday) lunchtime, at 11:30, directly in front of the
conference center (along the street), to find a place and discuss
architecture issues over lunch. Likely topics include layering model
stuff, negotiation, naming/address-neutrality in the transport layer,
etc. - it's open-ended at this point, but we'd at least like to
identify some of the most promising/urgent next steps. Since lunch
period is short, we'll leave promptly no later than 11:40, so be on
time. :)
Thanks,
Bryan
