RE: [PWE3] Proposed change to 5..4.1 of PWE3 architecture

["Yaakov Stein" <yaakov_s at rad.com>]

> > Following that, I suggest that we change the layering diagram such 
> > that the layers include only: Physical, Data-link, PSN, 
> > PW-Multiplexing, Payload Convergence, Payload (where payload 
> > convergence include timing, sequencing, etc but the decomposition is

> > not explicitly shown on the figure). The payload convergence
includes RTP and the control word.
> > 
> > We write a single architecture layer section which is good for both
IP 
> > and MPLS, since we don't specify the internal order of RTP and
Control 
> > word within the convergence layer in this section.
> > 
> > We write another section that describes the Convergence layer 
> > decomposition and specify (unfortunately) that the RTP and control 
> > word order changes depending on the demulplexing layer. If MPLS is 
> > used as demultiplexing layer RTP is after control word, else RTP 
> > before control word, or whatever)
> 
> How important are these changes?
> 
> If we make this level of change we really risk a significant 
> delay in getting the document back to the IESG, and since 
> this document gates all the encapsulation and control drafts 
> we delay them also.
> 
> Perhaps it would be better to rule Yaakov's proposal out of 
> scope for a draft that has already passed last call on this text?

While I think that Ron's suggestions 
(having first a common section, and thereafter separate IP and MPLS
ones)
are sensible, they seem to be basically editorial in nature.

At this late date I wouldn't want to hold everything up for
such editorial changes.

On the other hand, the use of UDP/IP has been in our
charter from the very beginning, even if most of the work
has been MPLS-oriented. 

I would therefore object to removing the use of pure UDP/IP
from the architecture document.

Y(J)S

_______________________________________________
pwe3 mailing list
pwe3@ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3