RE: [L1vpn] Proposed charter for L1VPN WG

["Hamid Ould-Brahim" <hbrahim at nortel.com>]

Loa,

(I understand this email wasn't intended specifically
to me) but you raised an interesting point I would
like to comment on it. 
> 
> Guess that Hamids comment boils down to that GMPLS can
> do more than L1.

Back like four years ago, some of us (ietf folks) proposed
the topic of "Optical VPNs" (GMPLS VPNs). So we had at that 
time requirement draft and some solution drafts here and there. 
The feedback from presentations and discussions at IETF 
meetings came that Optical is not really covering TDM. 

We went back and made it clear that it is about 
"Optical/TDM" VPNs (no change on technical content just title 
change). After other IETF meetings, the feedback came as well 
that Optical VPN is not really within the scope of IETF and 
a proposal was made to call it Generalized VPNs which are VPNs 
built using IETF GMPLS protocols and they apply where GMPLS 
apply. 

We went to the work we have done and made another minor 
change (pretty much find and replace function). Changes things like 
"Connection" to "LSPs", Optical/TDM to GVPN, etc. We didn't really 
change the functionality of the 
mechanisms which stayed pretty much the same.

That is to say that it is not a major work to consider
services other than L1 as long as functions are
well defined using GMPLS and VPN constructs.

But I am okay if we focus initially on L1VPN as a service,
my suggestion would be that from protocol point of view the work 
should aim when possible to use GMPLS (and VPN constructs) 
regardless whether they apply to L1 or other interfaces 
supported by GMPLS such that if later on we decide to extend the 
mandate of that group we already have most of key protocol
work done.


> 
> I don't think that this wg is the place to solve all genral VPN 
> problems, would rather urge to focus this really narrow e.g. the two 
> service models below
> 

I am fine with that. In fact I believe that the second
service model can be further subdivided in two
separate service models. One when the GMPLS provider network will 
appear to the client nodes as *one* private GMPLS-enabled node, 
and the other where the provider network appear as a subset of 
provider network resources (or nodes) to the client nodes as a  
private Gmpls network (fully interacting with client network).

Hamid.