[Nml-wg] Review deliverable 1

[Freek Dijkstra]

Below is a review by Franco Travostino (OGF area director) of
deliverable 1. I like to thank him again for making this review.


Franco has two remarks, and I propose the following.

I like to ask the authors of deliverable 1 to:
- Consider using consistent terminology for "hybrid network" or
 "optical network" in the title, as per Franco's comments.
- Correct/Add the URL as aksed by Pascale Primet on the list.


Immediate following these corrections, I will ask Gerg Newby to publish
the document as informational.


For the records, the authors are:
Paola Grosso, Anand Patil, Pascale Primet, Aurélien Cedeyn, Jason
Zurawski, Aaron Brown, Martin Swany, Freek Dijkstra, and Jeroen van der Ham.

Deliverable 1 can be found at: http://forge.gridforum.org/sf/go/doc14679


Furthermore, I like to point the list (in particular authors of
deliverable 2) to Franco's comments on the scope of this framework (on
path computation by application or human). Perhaps his insight is
beneficial while describing the schema we have.

Regards,
Freek


-------- Original Message --------
Subject: 	NML-WG, D1 comment
Date: 	Tue, 12 May 2009 22:36:02 -0700
From: 	Franco Travostino <travos at ieee.org>
To: 	Paola Grosso <grosso at science.uva.nl>, Richard Hughes-Jones
<Richard.Hughes-Jones at dante.org.uk>, Martin Swany <swany at cis.udel.edu>,
Freek Dijkstra <freek.dijsktra at sara.nl>, Cees De Laat
<delaat at science.uva.nl>

Dear All,
I re-read the draft with interest. And with a fresh mind too, given that
I've not spelled lightpaths in a really long while.

My chief observation is that the draft is not entirely clear on the
domain of application. At times, I feel that it's unduly overspecified.

The title indicates that it's for optical networks.
The intro talk about hybrid networks, which one could literally take to
mean heterogeneous network spans, whichever type they come in.
The examples talk about networks bearing some attributes.

Wouldn't this framework apply to any network with meaningful and diverse
enough attributes such that either a human being or an application will
want to do some dynamic path computation? I'm sorry that I cannot find a
shorter way to describe this ... but this is, I believe, a more generic
statement of purpose. I would pitch the framework in such generic terms
and then realize bits and pieces of it, so one embodiment will be all
about lightpaths, another one will be about some other circuit-switched
paradigm.

Your thoughts?

Any comment on this draft by the subject matter experts?

thanks
-franco

---
blog: http://www.thingsthatscale.com