RE: Comments on LMP draft version 04

["Bernstein, Greg" <GregB@ciena.com>]

Hmm, once again terminology does come up and the layering concepts prevalent
in the transport world.  DCC and various Jx bytes get associated with
certain layers and not just with some generic "data plane".

Innuendo's and lack of civility aside, the question does bear asking (does
LMP belong at the IETF) since the brunt of supporting LMP seems to be
falling on vendors of SDH/SONET equipment that already has built in support
for almost all the functions that LMP encompases.  This comes from lack of
scoping in the LMP specification and leads to customer confusion.  

Also once SDH/SONET overhead starts being "overloaded" this needs to be
coordinated with the appropriate bodies (ITU-T, T1X1).  When I brought up J0
based discovery back in 2000 at the OIF there were protests for this same
reason.  Now we have the G.7714 recomendation at the ITU-T (general
automated discovery) and a draft G.7714.1 on discovery for SDH and OTN.

What any of this has to do with IP is up to the ADs discretion I assume.  A
IP router could run LMP so it could be considered in scop.  But then all IP
routers with optical interfaces that I know of are SDH/SONET path
termination equipment does that put all of SDH in scope?

Greg

-----Original Message-----
From: Kireeti Kompella [mailto:kireeti@juniper.net]
Sent: Monday, August 05, 2002 2:51 PM
To: Bernstein, Greg
Cc: Michiel van Everdingen; Yakov Rekhter; Jonathan Lang;
ccamp@ops.ietf.org
Subject: RE: Comments on LMP draft version 04


Hi Greg,

On Mon, 5 Aug 2002, Bernstein, Greg wrote:

> IP in the data plane for test messages?  Kireeti most optical circuit
switch
> equipment (switches) can't touch the IP in the data plane and wouldn't
even
> know that its carrying IP.

Once the circuit is set up, a switch would not need to know what it
is carrying.  However, Test Messages are sent *before* set up.  In
this phase, many switches are capable of terminating enough in the
"data plane" to allow Test Messages to be used.

I think the problem is again one of terminology.  You may not consider
DCC or J0 part of the data plane.  However, they are often inextricably
linked with the data plane so that identifying which fiber/port/lambda
a given DCC channel or J0 string is on automatically identifies which
fiber/port/lambda the data link is on.  That is the entire intent of
the Test Message.

> What are you getting at below with you're comment?  Note that J0, J1 or J2
> strings are not even communications channels and hence you couldn't send
IP
> test messages.  Line and Section DCC are not considered data plane but
> overhead in SDH/SONET.  Hmm, does LMP not belong in the IETF?

Perhaps it is a mistake to believe that this issue can be solved in a
civil and especially in a technical manner.  Innuendos like your last
Hmm comment are not in the least helpful, and many might even think
that they were a deliberate attempt to sabotage the progress of the LMP
in the IETF.

What I'm getting at with my comment is to reiterate the IETF's goals
in this (and all other) matter(s), namely, to further the progress
of the Internet Protocol and the Internet.  This issue is particularly
sensitive in the WGs in the Sub-IP area.  Protocols like GMPLS and LMP
may well find utility in non-IP networks; however, that is a far cry
from mandating that a protocol must work in a non-IP environment.

Now, can we get back to technical matters?

Kireeti.