RE: LMP & neighbor discovery

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

Of course I comment on text I don't read :-). I get down load the updates
check the abstract, TOC intro for changes and if I don't see something that
impacts me there I don't dig into the details looking for new features...

But I'm still confused about whether "automatic discovery" is in the spec.
Michiel seems to have done a lot more complete recent scrubing and can't
find it.  It's standard in OSPF and IS-IS to be able to discover neighbors.
With the OIF UNI we added this for SONET/SDH in the "in-band" case.
I'm assuming any vendor that has deployed a fair amount of optical control
plane enable equipment has put in discovery to avoid the need to "configure"
TE links.  The carriers I know "wouldn't leave home without it".

Since it wasn't in LMP, this was taken up by ITU for the SDH/G.709 case with
the work in progress on G.7714.1.  

One issue does cross my mind on control channel management is why not use
something like SCTP?  It seems to have nice redundancy capabilities, assured
delivery, etc...  A portion of its functionality could just be used.  

Greg

***********************************
Dr. Greg M. Bernstein
Senior Technology Director, Ciena Corporation 



-----Original Message-----
From: Yakov Rekhter [mailto:yakov@juniper.net]
Sent: Friday, May 31, 2002 2:29 PM
To: Bernstein, Greg
Cc: 'Jonathan Lang'; 'Martin Dubuc'; Michiel van Everdingen; Kireeti
Kompella; ccamp@ops.ietf.org
Subject: Re: LMP & neighbor discovery 


Greg,

> Hmm, Jonathan it seems the spec never stops changing.  It seems to have
been
> a "one size fits all" as long as its my flavor of PXC.  Am I supposed to
> implement just part of the spec in the case of SONET/SDH? 

Yes.

> How does a carrier specify this?

I am sure carriers can figure this out. After all, LMP is not
the only protocol that has optional parts.

> It seems better if it was broken up in multiple specs or you get very
> specific concerning the technology you are addressing.  It
> doesn't make sense for a technology that already provides fault management

> and performance monitoring to repeat this functionality.

I certainly agree with the point you made in the last sentence. And
that is why fault management functionality in LMP is optional.

> So you're now telling me that after constantly fighting me to keep "true
> discovery" out of LMP, you've now put it in?  Can you update the
> introduction to the draft to reflect this.  Most of us don't have the time
> to read through the details after the intro portion says it
> doesn't have the functionality we desire.

Are you trying to say that you commented on the text that you didn't
even read ?

Yakov.