RE: [PWE3] PWE3 new WG items?

[benjamin.niven-jenkins at bt.com]

Peter,

> I agree that the PWE3 WG could have decided to treat PWs as 
> LSPs (which currently is not true) and to use RSVP as the 
> protocol for PW set up.

The PWE3 WG may have decided not to treat PWs as LSPs but it doesn't (IMO) avoid the fact that a PW *always* uses a LSP for transport.  AIUI a PW consists of two parts:

1) An adaptation function that describes how to encapsulate a client layer (e.g. Ethernet) onto a server layer (e.g. MPLS). This is defined by the PWE WG.

2) A LSP (let's call it the 'PW LSP') that transports the above adaptation function (and its client payload) across a network.  This is defined by the MPLS WG with some PW specific extensions (e.g. PW FEC)

For scalability etc. a second LSP (let's call it the 'transport LSP') is also used and a 2 level LSP hierarchy/stack is created with the inner LSP being the 'PW LSP' (which is a single hop) and the outer LSP being the 'transport LSP'.

Now my argument is that any QoS parameters associated with a PW are in actual fact associated with the 'PW LSP' and not the adaptation function.  Therefore any QoS parameters/signalling that are associated with a PW are in actual fact associated with the 'PW LSP'.

Therefore I believe that the signalling of PW QoS parameters is the same problem as the signalling of LSP QoS parameters and therefore does not fall under the PWE3 WG.




_______________________________________________
pwe3 mailing list
pwe3@ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3