The MPLS WG Archive

["Hamid Ould-Brahim" <hbrahim@nortelnetworks.com>]

Richard,

When deployed over an MPLS infrastructure the VR model
can use a two label approach to forward vpn packets 
transparently over the service provider network. 

Depending on the MPLS deployment scenario, 
the inner label can be bound to the VR or to 
the vpn prefix within the VR context. In the latter case, 
the VR is responsible to distribute the vpn labels attached
to the private reachability information. Each VR may or may
not run MPLS. Both schemes can also be used. As an example,
the per vpn label distribution can be tunneled (with MPLS or not),
where the inner label is bound to the VR. Once it is distributed
to each VPN, the data path will just use the two labels, one
for the vpn prefix (attached in VR context), and the top label
within the PE (or backbone VR). Because each VR is independent
from the others, some VRs may be running MPLS, some may use 
IPSec, etc.

Regards
Hamid  

> 
> "Morgan, Richard" wrote:
> >
> > Does the Virtual Router soluntion take advantage of the 'hierrachy of
> > routing knowledge' ie use two labels, one to get the PE and another to
> > identify the VR - or does it create individual LSP's between each other VR's
> > and not use the label stack ?
> >
> > Richard Morgan
> >