Re: [RAM] Crisper Terminology

On 06/09/2007 02:42 AM, Brian E Carpenter allegedly wrote:
> I'd be tempted to agree with that, but I just had occasion to
> review draft-ietf-ccamp-gmpls-addressing-07.txt, and found my head
> spinning rapidly on exactly this question.

The problem there is separation of planes.  Router and Forwarder are
somewhat independent, thus Control and Forwarding (Data) Plane are as
well.  In GMPLS you have an independent control plane, and an
independent control plane requires independent transport names.  The
IP component of control plane transport, has its own locators, and
forwards packets which carry in them identifiers of components of the
(possibly non-IP) forwarding plane.  I believe it's right to say that
the "addresses" in the forwarding plane are really identifiers, but it
depends how path setups are done.  However, it seems that those
identifiers are not only system identifiers but also interface or link
identifiers.

> Well, if we use OSI formalism, we'd have to say that the transport
> address is the network address concatenated with the port number.
> There was a fairly clear model for layered endpoint addressing in
> OSI, and it still a valid thinking aid IMHO.
> 
>     Brian
> 
> _______________________________________________
> RAM mailing list
> RAM@iab.org
> https://www1.ietf.org/mailman/listinfo/ram
> 

_______________________________________________
RAM mailing list
RAM@iab.org
https://www1.ietf.org/mailman/listinfo/ram