RE: Draft charter for L2VPN - IPLS & ARP Mediation

On Thu, 2003-05-15 at 20:47, Hamid Ould-Brahim wrote:
> Giles,
> 
> > 
> > IPLS meets the requirement for which I originally envisaged VPLS being
> > useful (i.e. interconnecting IP routers or hosts) and does this more
> > efficiently than VPLS.  I would be reluctant to use VPLS to 
> > interconnect
> > switches, as that exposes the SP to the customer network's MAC
> > addresses, but I can see why this might be attractive to customers in
> > the metropolitan area at least.
> > 
> 
> What do you suggest as a service to interconnect switches
> besides using VPLS? (particularly across IP/MPLS network).

I don't think there is much alternative to VPLS for this (other than
using meshes of draft-martini VCs, and either running spanning tree over
the top or implementing split horizon in the CE devices).

> (your observation seems similar to l3vpns where the SP is
> exposed to customer routes which seems okay for some 
> customers)...

Agreed.  But of course there tend to be more MAC addresses than prefixes
in the average network.

And, just to be pedantic, the scaling problem with 2547 or VPLS isn't a
question of what the customer is okay with, but rather of what the
provider can support.

Giles

-- 
=================================================================
Giles Heron    Principal Network Architect    PacketExchange Ltd.
ph: +44 7880 506185              "if you build it they will yawn"
=================================================================