- A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
RE: L3 VPN/Route reflector architecture [9:2119] posted 09/27/2002
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]

But I will be advertising the BGP routes to the rest of the network
through regular BGP.  The core will only see a route for x.x.x.x/x with
the next-hop as one of my L3 VPN routers and they will take it from
there.  Also, there will also be no overlapping address spaces, I do not
plan to advertise private subnets across the core.  The only thing the
core will need to know is how to reach the public subnets, of which none
will be overlapping.


-----Original Message-----
From: alok [mailto:alok.dube@xxxxxxxxx] 
Sent: Friday, September 27, 2002 1:13 AM
To: Eric Van Tol; juniper@xxxxxxxxxxxxxx
Subject: Re: L3 VPN/Route reflector architecture [9:2119]
Importance: High

Hi Eric,

i just re read ur question.....

it wont work...coz u wont be able to distiguish between the overlapping
address spaces and hence the "VPN" to which their packets belong...
the RD is used to distinguish route, the MPLS header to distinguish
VPN the packet belongs too... so the header is needed...hence u need
the core...


----- Original Message -----
From: alok 
To: Eric Van Tol ; 
Sent: Thursday, September 26, 2002 11:05 PM
Subject: Re: L3 VPN/Route reflector architecture [9:2119]

i think for BGP MPLS VPNS too...if route reflectors are defined (need to
check that 1st) they wud be on a per "VPN" bases...
so u cant say the route reflector would reflect routes across BGP peers
different VPNs...

i guess some one cud confirm this... but thats what i think shud be the

but b) is fine, u may not need BGP in the core... i dont see how it wud
matter if u didnt have it..
----- Original Message -----
From: Eric Van Tol 
Sent: Thursday, September 26, 2002 7:08 PM
Subject: L3 VPN/Route reflector architecture [9:2119]

Hi all,
Let's say I have a network with 5 routers:

P-core-A P-core-B
    |     |
    |     |
         PE-A ------PE-B------PE-C

The two core routers are route reflectors for the three PE routers.  No
iBGP peering happens between the PE routers.  The three PE routers can
also reach each other directly, as they are all on the same subnet.

Now I introduce the MPLS VPN and it's necessity to pass routing
information for VPNs through BGP.  Can I set up the iBGP peering
sessions for just the VPNs between the three PE routers without having
to a) set up MPLS on the core routers, and b) set up VPN iBGP peering on
the core routers.

Since the only information that is passing between the PE router iBGP
sessions are the VRF tables, I would think that this type of setup is
fine.  Am I mistaken?


Message Posted at:
FAQ, list archives, and subscription info: