Re: [CCAMP] Resource, Resource Block, and Resource Pool in draft-ietf-ccamp-rwa-wson-encode and draft-ietf-ccamp-rwa-info

[Leeyoung <[e-mail]>]

Hi Adrian,

I think we are on the same page. Please see inline for specific comment. 

Best regards,
Young

-----Original Message-----
From: Adrian Farrel [mailto:[e-mail]] 
Sent: Friday, January 23, 2015 9:27 AM
To: Leeyoung; [e-mail]
Cc: [e-mail]; [e-mail]
Subject: Resource, Resource Block, and Resource Pool in draft-ietf-ccamp-rwa-wson-encode and draft-ietf-ccamp-rwa-info

Hello,

Continuing this discussion, I think we are getting closer.

> > My comment for general vs. WSON-specific issue is that  the design
philosophy
> > of general encoding was to generalize a common element that can be applied
> > to different technologies. For instance, connectivity matrix definitely can
be
> > applied to WSON, OTN and other Switching technology as one encoding can
> > fit to all. You seemed to desire a sharing of the same encoding to describe
> > different entities where possible. This is fine for label vs. wavelength as
> > there is one-to-one mapping for this. However, Resource Block encoding
> > cannot properly share with the connectivity matrix encoding with two
> > reasons:
> >
> > 1. They refer to different elements in the node.
> > 2. They require different sets of fields with different number of fields.

OK, what we really seem to be running into is some rather vague definitions of
"resource" and "resource block".
Do you think we could spend some time nailing those down. Then we'll work out
where to put them (possibly in rwa-info).

YOUNG>> Yes, I have just noticed that you held rwa-info and I hope you can suggest some clarifying text into that draft, which I see below. 

The question is not about the protocol element "resource block" but is about
what it represents.

So, as a starter, what is a Resource in a WSON system?
You have said (lower down this email) that " resources meant
regenerators/wavelength converters".
I can live with this (although it is a long way from the term "resource" used in
3473 etc. where it is taken to mean buffers, bandwidth, memory, labels,
lambdas,...
That difference in interpretation is sufficiently large that it needs to be
brought out and made very clear.
You need something like:
   In this document the term "Resource" is used to refer to a
   physical component of a WSON node such as a regenerator
   or a wavelength converter. Multiple instances of such
   components are often present within a single WSON node.
   This term is not to be confused with the concept of
   forwarding or switching resources such as  bandwidth or
   lambdas.

YOUNG>> Great suggestion. Yes, that is what is Resource is. The part of the problem has been that the term 'resource' was picked when the WG asked us to generalize Regenerators/wavelength converters. This term 'resource' may have not been perfect, but it was chosen as such to have its meaning in a specific context. 

Then you can answer what is a Resource Block in a WSON system?
>From what I think I now understand, a resource block is simply a collection of
resources on a WSON network node that behave in the same way. This allows easier
description in the protocol encoding, but has no other meaning. So you could
say:
   A Resource Block is a collection of Resources from the same
   WSON node that are grouped together for administrative reasons
   and for ease of encoding in the protocols. All Resources in the
   same Resource Block behave in the same way and have similar
   characteristics relevant to the optical system, e.g., processing
   properties, accessibility, etc. 

YOUNG>> Yes, it is correct.

Then comes, what is a Resource Pool? Actually, this is only implicitly defined
in draft-ietf-ccamp-rwa-info and so we have quite a gap. It looks like you might
say:
   A Resource Pool is a collection of Resource Blocks for the
   purpose of representing throughput or cross-connect
   capabilities in a WSON node. A Resource Pool associates
   input ports or links on the node with output ports or
   links and is used to indicate how signals may be passed
   from an input port or link to an output port or link by
   way of a Resource Block (in other words, by way of a
   Resource). A Resource Pool may, therefore, be
   modelled as a matrix.
   
   A Resource Block may be present in multiple Resource
   Pools.

YOUNG>> Yes.

And finally, we have Resource Block Set. What is that?
I *think* it is just the encoding concept for a Resource Block Pool.
You have (in 2.1)
   In a WSON node that includes resource blocks (RB), denoting subsets
   of these blocks allows one to efficiently describe common properties
   of the blocks and to describe the structure and characteristics, if
   non-trivial, of the resource pool.
*Now* I see that your "subsets" should actually be "sets" and that will make
more sense.

YOUNG>> Yes, "sets" make it clearer.

But do you need the two terms "Resource Block Pool" and "Resource Block Set"?
Are they the same or different? When I read 3.2 it looks like the Resource Block
Pool is a collection of Resource Block Sets.

YOUNG>> They are different. Resource Block Pool is sitting above the RB Sets. There may be multiple RB Sets in a node that behave differently, e.g., different input/output constraints on the access/egress to the Pool. 


Have I finally got this right?

YOUNG>> Yes, I think so. 

If so, then can we agree precise text and where to put it?

YOUNG>> Yes. 

Thanks for your patience.
Adrian