RE: Optical Link Interface - LMP-WDM comments

[Bala Rajagopalan <BRaja@tellium.com>]

Jonathan & Andre:

My intent was to give a constructive suggestion
on moving forward. Specifically, if the motivation
for NTIP was to do a targetted protocol for OLI,
the same purpose could perhaps be achieved by
streamlining LMP-WDM. So, please look at the technical
issues pointed out and of course, we can discuss these.
Some of my questions in the attachment were for clarification
also. 

With regard to Jonathan's comments:

1. OLI: Yes, I am an author in the OLI requirements. I have 
repeatedly said that OLI is important. Yet, I don't (and didn't)
agree with every requirement in the OLI document. (I'm not
the only author in that document). The way I see it, we
just have the 00 version. Things could change going forward.

2. LMP vs LMP-WDM: I've indicated some particulars about this
in the attachment. I'll look into this further.

3. Link verification: It's a matter of cost vs benefit.  Again,
as I said earlier, this is an area requiring more thought and
refinement.

Regards,

Bala


-----Original Message-----
From: Jonathan Lang
To: Jonathan Lang; ''Bala Rajagopalan ' '; ''Andre Fredette ' '
Cc: '''ccamp@ops.ietf.org' ' '
Sent: 7/30/01 5:37 PM
Subject: RE: Optical Link Interface - LMP-WDM comments

Bala,

[replying to my previous email]

I just realized there was an attachment to your email.  Having gone over
those details, I see you have given specifics to your points (1) and
(4).
Please ignore my comment for (4) - we will work on clarifications here.
I
believe my comments on the other points, other than "What specific
information are you talking about here?" in (1), are still valid.

Thanks,
Jonathan

-----Original Message-----
From: Jonathan Lang
To: 'Bala Rajagopalan '; 'Andre Fredette '
Cc: ''ccamp@ops.ietf.org' '
Sent: 7/30/2001 1:47 PM
Subject: RE: Optical Link Interface - LMP-WDM comments

 Bala,
  Please see inline.

Thanks,
Jonathan 

>-----Original Message-----
>From: Bala Rajagopalan
>To: Andre Fredette
>Cc: 'ccamp@ops.ietf.org'
>Sent: 7/30/2001 7:38 AM
>Subject: Optical Link Interface - LMP-WDM comments
>
>Andre:
>
>I'm attaching specific comments on LMP-WDM. Top level issues are
>as follows:
>
>1. Layer mix-up: In some places, the WDM system is required to
>send information which is not native to OLS to the OXCs. In fact, 
>this is very strange, because this information must originate from
>the OXC domain, get configured in WDM, and then go back to OXCs. This
>is a problem with OLI definition, not LMP-WDM itself. This requires
>some discussion and clean up. 
What specific information are you talking about here?  As you say, this
is
an issue with OLI definition itself and the parameters were discussed at
quite length among both LMP-WDM authors and NTIP authors.  (I do notice
that
you are also listed as an OLI author...).  Certainly we can revisit
these
issues if they need to be addressed...

>2. Rather than tailoring the procedures for the particular requirements
>(OLI), the draft seems to be written with the goal of maximal reuse of
>LMP. This sometimes adds to the complexity of the proposal.  
LMP-WDM was written to satisfy the OLI requirements and reuse the LMP
protocol.  This is similar to the approach taken for GMPLS RSVP-TE and
CR-LDP.

>Specifically, LMP-WDM has many "extensions" to LMP, giving the
impression
>that LMP-WDM is a superset of LMP. I think there are differences in the
>situations where LMP is applied as compared to LMP-WDM and these
aspects
>are not discussed sufficiently. I think a meticulous examination of the
>OLI requirements and LMP procedures is necessary to just carve out a
set 
>of procedures for OLI, rather than the superset approach. The maximal
>"re-use" argument doesn't make any sense on the WDM side where there is
>no LMP. Or, for that matter, where OXCs don't implement LMP. (This is
>where NTIP makes a good point).
Which LMP-WDM procedures are not required for OLI?

>3. I'm afraid the current LMP link verification definition (and as a
>corollary, LMP-WDM link verification) has only a marginal value
>considering the investment on hardware and protocol implementation.
This 
>is an area that requires more work.
If you believe link verification has marginal value, you don't have to
implement it (remember it's optional).  We have heard otherwise from
customers.

>4. Some of the procedures in the draft must be more explicitly
described
>rather than saying that they are same as LMP's.
Please clarify which procedures you are referring to.

>As you noted, a more careful examination of LMP itself is in order.
This is greatly encouraged.

I also think it would be useful for you to do a similar analysis of
NTIP.

Thanks,
Jonathan

>Regards,
>
>Bala
>
>Bala Rajagopalan
>Tellium, Inc.
>2 Crescent Place
>P.O. Box 901
>Oceanport, NJ 07757-0901
>Tel: (732) 923-4237
>Fax: (732) 923-9804
>Email: braja@tellium.com