Re: [PWE3] SG15Q12 liaison on MS-PW requirements

["Francesco Fondelli" <francesco.fondelli at gmail.com>]

On Nov 22, 2007 7:18 PM, Stewart Bryant <stbryant at cisco.com> wrote:
[cut]
> picked up on) that there is only ONE label
> in the packet that does double duty between PW label
[cut]

No, IMHO that statement says something different.  G.805
terminology/methodology is
the whole point here. Even if I am far to fully understand G.805
functional architecture
I try to explain what the liaison says (at least my interpretation).
We have to agree
on few points before any dissertation:

1)
In figure 1 (liaison 2007-09-20) "T-MPLS path" *functionally matches*
a plain LSP.
This entity is just *described*, from a functional view point, in G.8110.1
(using G.805 terminology/methodology).

Please note that the entire "T-MPLS thing" can be defined as -- a *description*
of MPLS from a different view point + some additional constraints (no PHP, only
connection-oriented, etc) + a specific OAM mechanism --. I understand
this might sound strange.

2)
In figure 1 (liaison 2007-09-20) "T-MPLS channel" *functionally
matches* a plain PW.
It is so close to a PW that I call it "PW" :-)

3)
In G.805 formalism a triangle is a "trail termination" and a trapezium
is an "adaptation",
again we are talking about *functional blocks*.  If you want to
describe a network
architecture in G.805 terms you begin describing each single network
element functional
block: you'll describe what goes in ("input signal"), what comes out
("output signal") and
what a single bloody block in the middle should do ("functional block").
ITU-T people claim they can describe everything using this paradigm (SDH, MPLS,
even IP or PBB or whatever).

4)
Figure I.3 of G.8110.1 illustrates how the PWE3 concept fits within
the T-MPLS layer network model (again this is a *description*) and Figure I.4
(same rec, G.8110.1) illustrates how a MS-PW can be *described* using
the same framework.
Please note that ITU-T people say that an Appendix is just informative...
... hemmm actually is the only part of a ITU-T recommendation that I
vaguely understand.

5)
A G.805 trapezium functional block usually performs mux/demux
operations. This is
graphically represented drawing more then one incoming line (5.3.3.1.1, G.805).
However the TMC/PW trapezium (liaison fig 2) has *only one* ingress line
(in G.805 terms I guess we can say that it accepts only one client signal).

----------

Ok. If someone doesn't agree with me about above statements please
*stop reading*
and try to provide alternative interpretations.

The liaison, in my opinion, says: "We (T-MPLS guys) already have a functionally
equivalent PW in our model (i.e. T-MPLS channel trail). If we attach
this *thing*
to a PW segment we obtain a MS-PW.  Do you mind if we do something like that?".

> there is only ONE label in the packet

there are TWO labels in that packet, because the TMC/PW trapezium usually
pushes one specified new label onto the label stack (somewhere in G.8110.1 or
G.8110).

I beg the "T-MPLS design team" to produce an informational I-D to
address these terminology issues (something like rfc 4397 for
GMPLS-ASON) otherwise
this mess will never stop.  Moreover I suggest that people
understanding G.805 functional
architecture stand up and correct me about what I have just said.  I
really need someone
telling me whether I'm right or not!

>
> - Stewart

Ciao, my 2 cents
FF

ps
in its infinite wisdom ITU-T has made free/available all recommendations
referenced in this email (http://www.itu.int/rec/T-REC/e), I have no access
to new unpublished amendments.


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