The MPLS WG Archive

[Javier Antich <javier.antich@telindus.es>]

One of the main advantages of using MPLS in the core is to avoid running BGP
in the core routers, as long as they only switch packets (except from those
for IGP routing advertisments) but not route them. Of course, if the MPLS
cloud fails due to, for example, a software bug, the core routers would need
to route packets according to their routing tables. You could have the core
routers running BGP and MPLS, but most of the time BGP routes would not be
used at all. If the MPLS code fails, I am not sure whether your services
(like VPNs, Traffic Engineered Paths, ...) would go on working, even if you
have BGP running in your core routers.



Visite nuestro web: http://www.kerndatanet.com/
<http://www.kerndatanet.com/> 
                               http://www.telindus.es
_________________________________________________________________________
Javier Antich Romaguera.                          KERN DATANET -TELINDUS
ESPAÑA
Dpto Pre-Venta.	                                       Torre Metropolitana

                                                              Plaza Ciudad
de Viena, 6 -2º 
                                                              28040 MADRID.
(ESPAÑA).
e-mail: javier.antich@telindus.es <mailto:jantich@kerndatanet.com>
TF:  (+34) 91 4560008




	-----Mensaje original-----
	De:	Chris Flores [SMTP:chris.flores@onfiber.com]
	Enviado el:	jueves 28 de septiembre de 2000 20:18
	Para:	'Javier Antich'; Michel Redondo Ferrero
	CC:	mpls@UU.NET
	Asunto:	RE: MPLS/BGP routing question

	Interesting, then I have the following question. Let's say the
transit
	backbone consists of a 3 level hierarchy - core, distribution and
access.
	BGP is configured such that the access (or edge) routers are route
reflector
	clients of the distribution routers. Furthermore, the distribution
routers
	are route reflector clients of the core routers. As Michel Redondo
Ferrero
	has stated, MPLS VPNs originate and terminate on the access or edge
of the
	network (in his scenario). Why would you turn off BGP on the core
routers?
	What if MPLS breaks or fails for any reason (i.e. software bug).
Then, how
	would routing occur?

	Regards.

	chris

	-----Original Message-----
	From: Javier Antich [mailto:javier.antich@telindus.es]
	Sent: Thursday, September 28, 2000 11:28 AM
	To: Michel Redondo Ferrero
	Cc: mpls@UU.NET
	Subject: RE: MPLS/BGP routing question


	Core Routers do not need to run BGP because they, in practice don't
take
	routing decisions with user traffic (I mean traffic comming or going
to
	Internet), they just switch frames based on their MPLS label. BGP
decissions
	are made at the edge and packets sent to the BGP next hop through
the
	corresponding MPLS LSP.





	Visite nuestro web: http://www.kerndatanet.com/
	<http://www.kerndatanet.com/> 
	                               http://www.telindus.es
	
_________________________________________________________________________
	Javier Antich Romaguera.                          KERN DATANET
-TELINDUS
	ESPAÑA
	Dpto Pre-Venta.	                                       Torre
Metropolitana

	                                                              Plaza
Ciudad
	de Viena, 6 -2º 
	                                                              28040
MADRID.
	(ESPAÑA).
	e-mail: javier.antich@telindus.es <mailto:jantich@kerndatanet.com>
	TF:  (+34) 91 4560008




		-----Mensaje original-----
		De:	Michel Redondo Ferrero [SMTP:mredondo@idecnet.com]
		Enviado el:	jueves 28 de septiembre de 2000 8:26
		Para:	mpls@UU.NET
		Asunto:	MPLS/BGP routing question

		Hi,

		Considering the next scenario:

		-Core and Border routers running IS-IS, MPLS
		-VPNs configured in Border routers using BGP/MPLS
		-Border routers running BGP with full-routing

		The question:
		Do Core routers need to run BGP? Is IS-IS enough?

		Thanks in advance for your answers.

		Michel Redondo Ferrero