Re: VPLS model for L2VPN Framework document

At 01:27 PM 5/30/2003 +0900, Muneyoshi Suzuki wrote:

> > The  way the  framework document  is written,  it seems  to 
> suggest  that an
> > "emulated LAN"  is the as a "VPLS  instance", which in turn  suggests 
> that a
> > single  MSTP instance  cannot govern  multiple  VPLS instances.   So 
> I  just
> > wanted to  make it clearer that  an emulated LAN  could consist of 
> a  set of
> > VPLS instances plus  an MSTP instance; I think this  would remove 
> Mick's and
> > Norm's objection.
>
>It seems to me, something is still confusing. What is different between
>emulated LAN and VPLS instance? Could you clarify this?

This is the way I look at it:
A VPLS instance corresponds to an end-to-end service instance that can 
spans across different access networks (e.g., one side to be QinQ and the 
other side to be MPLS).

An emulated LAN correspond to a set of PWs and their associated forwarding 
functions.

An emulated LAN looks like an emulated VLAN with respect to the bridge module.


>Regarding the framework model, if a single PE contains multiple logical
>Bridges, a single VPLS Forwarder cannot be shared by Bridges. In this
>case, a PE must contain multiple logical Bridges and VPLS Forwarders. Each
>Bridge may have a single MSTP instance, and the instances interwork through
>VPLS Forwarders. I can't see any problem in this model. (Needless to say,
>if the VPLS Forwarders is supported by full mesh, there is a big issue ;-)

Yes, in that model there is a one to one correspondence between logical 
bridge and forwarder and sometimes the combination of the two is referred 
to as VSI.

-Ali


>Thanks,
>
>
>Muneyoshi Suzuki
>Nippon Telegraph and Telephone Corp.