The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2001-Feb> msg00137



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: REG: Label Distribution.

  • From: jim guichard <jguichar@cisco.com>
  • Date: Thu, 22 Feb 2001 19:23:46 +0000
  • Cc: Vinod Anthony Joseph Cherunni <vac@dsqworld.com>, mpls-ops@mplsrc.com
  • Resent-Date: Thu, 22 Feb 2001 15:45:05 -0500
  • To: "Ta,Phong G." <pta@mitre.org>
  • X-Sender: jguichar@london.cisco.com

you need to run a full mesh of MP-BGP sessions between PE routers that wish 
to receive routing information for particular VPNs - this means that it is 
not necessary to run a full mesh between ALL PE routers - you could run 
several meshes and split the topology. In most deployments, route 
reflectors will be used and PE routers will peer with whichever route 
reflectors hold the routing information they are interested in. Some 
further comments below .. Jim

At 12:04 22/02/2001 -0500, Ta,Phong G. wrote:
>Yes, you will have to do a full mesh of MP-iBGP sesions between the
>interested vrf VPN sites.  If the number of MP-iBGP is large, router
>reflector will help the configuration task; but the number of MP-iBGP
>still remains the same.
>
>Regards,
>
>Phong
>
>Vinod Anthony Joseph Cherunni wrote:
> >
> > Hi,
> >
> > Thanks for the inputs, Sorry if I am asking a very basic question. Is
> > it that even the routers that do have any CE routers terminated on
> > them locally & are part of the core, need to be aware of all the MPLS
> > VPN customer prefixes.

yes. If the PE does not have any attached VPN sites then it does not need 
to receive any VPN routing information from across the core.

>If so I would need to have a full mesh of
> > MP-iBGP in my core, Kindly correct me if I am wrong.

you do not need a full mesh unless ALL of your PE routers want ALL of the 
VPN routing information for every attached site.

>So the config
> > would be like -
> >
> > int s2/0
> >  ip vrf forwarding VPN1
> >
> > router bgp 1
> >  address-family ipv4 vrf VPN1
> >                 neighbor r1 remote-as my-as
> >                 neighbor r2 remote-as my-as
> >                  ....
> >  address-family ipv4 vrf VPN2
> >                 neighbor r1 remote-as my-as
> >                 neighbor r2 remote-as my-as
> >                  ....
> >

no. The ipv4 address-family is for eBGP connectivity to attached CE sites. 
To configure your interior BGP sessions, you must activate the neighbors 
within the vpnv4 address-family e.g

router bgp 1
  neighbor x.x.x.x remote-as 1
  neighbor x.x.x.x update-source loopback0
!
  address-family vpnv4
     neighbor x.x.x.x activate

> >   exit-address family
> >
> > So I would need to define my MP-iBGP neighbors once per VPN as given
> > above.
> >

no. You only need define your MP-BGP neighbors once, under the vpnv4 
address-family.

> > Kindly enlighten me.
> >
> > With warm regards,
> > Vinod.
>
>--
>Phong Ta                        The MITRE Corporation
>Phone:  703-883-6787            1820 Dolley Madison Blvd., MS:W650
>Fax:    703-883-7142            McLean, VA  22102
>
>-------
>The MPLS-OPS Mailing List
>Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
>Archive: http://www.mplsrc.com/mpls-ops_archive.shtml



Jim Guichard CCIE #2069
Technical Advisory Consultant EMEA

+44 208 756 8806
Mobile: +44 7802 809763

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml