Re: [dcp] Re: draft DCP charter for discussion

Craig -

>Well help me out here.  It seems to me that the protocol, as described,
>is an oxymoron.  I.e., without a closed loop there's no flow control
>and if the protocol is a true unreliable packet flow (e.g., UDP) there's
>no loop.
>
>So, I ask, is the plan for UDP with acks?  (I.e., the packet may not get
>there, but if it does, the sender hears about it).  Or are we talking
>something wilder?

There are three separate drafts, along with the main DCP document,
that specify congestion control profiles.  They are available on the
DCP web page at:
 http://www.aciri.org/kohler/dcp/
They are a large part of the point of DCP in the first place.

The TCP-like congestion control profile includes acknowledgements like
those in TCP.

The TFRC congestion control profile includes the kind of feedback in
TFRC.

The application gets to chose which congestion control profile it
prefers.

The application does *not* get to choose to send at a high rate
with no congestion control at all.

- Sally

_______________________________________________
dcp mailing list
dcp@ietf.org
https://www1.ietf.org/mailman/listinfo/dcp