Skip navigation
Cisco Learning Home > CCNP SP Study Group > Discussions
1291 Views 6 Replies Latest reply: Apr 22, 2013 2:13 AM by Amr M. el Adel-CCIE(RS,SP) RSS

Currently Being Moderated

BGP configuration

Nov 28, 2012 3:01 PM

Paul 13 posts since
Oct 30, 2012

Im having a bit of a problem understanding a BGP setup.  I set up a lab with 4 core routers.  They are all running OSPF.  The 2 edge routers are connected to 2 additional routers.  These additional routers have their own AS each.  I have them connected with eBGP.  The 2 edge routers in the core are connected to each other using iBGP.  My question is on the routers that are on their own AS that connect to the edge routers at the CORE.  Do i just need to put a network statement command on the router BGP configuration or do I need to redistribute the OSPF into BGP as well? 

  • David Flores 1 posts since
    Feb 13, 2010
    Currently Being Moderated
    1. Nov 28, 2012 3:43 PM (in response to Paul)
    Re: BGP configuration

    Hi Paul,

     

    Well it depends wich routes you are delivering to these eBGP routers, and actually what routes you have in your BGP table on your CORE routers. Setting up a BGP connection doesn't make the networks advertised by the protocol, you have to explicitly configure how to do this. The network statement is a good way to go, the redistribution not so much, but of course it depends on what your topolgy looks like and what you are trying to accomplish.

     

    Maybe if you post some configs we can help out a little more.

  • amarhold 53 posts since
    Nov 2, 2009
    Currently Being Moderated
    3. Dec 4, 2012 11:39 PM (in response to Paul)
    Re: BGP configuration

    Hi !

    I do not exactly understand what you trying to achieve BUT:

    regarding the MB-routers

    1. in OSPF passive-interface default means that you are not exchanging any OSPF with any other router and thus your OSPF will be empty ( no OSPF routes when you do a show ip route)

    2. in BGP when you specify a network statement with a mask you need an EXACT match of this route in the routing table which is not the case as the same network is a /32 on the loopback interface.

    So you need a static route pointing to Null0 to get the network announced by BGP

    ip route 172.22.208.0 255.255.248.0 Null0 250

     

    in your core you have enable LDP only on one link ?!

     

    hope that brings you a step further

    regards

    alexander

  • blau 58 posts since
    Oct 20, 2012
    Currently Being Moderated
    5. Jan 4, 2013 1:37 AM (in response to Paul)
    Re: BGP configuration

    Hi Paul

     

    From topology which you provided it seems that there is no direct connection between PE routers. It means running iBGP session via P routers which will work just fine but P routers will drop all traffic between MB routers cause they are unaware of their networks.

     

    One way to solve this problem could be redistribution between BGP and OSPF but it is not recommended.

    Here is list of other possible solutions:

     

    - full mesh iBGP sessions, which means P routers need to run BGP, but this is also not very recommended. P routers are supposed to be BGP free.

     

    - route-reflector/confederation, but again P routers will need to run BGP

     

    - create direct connection between PE routers, this will work only if direct connection will be functional. If it fail, iBGP session will reroute via P routers so no problem there. BUT also traffic between MB routers will be rerouted via P routers and they will drop it.

     

    - create tunnel (IPv4, GRE) between PE routers: P routers are BGP free, external traffic will be forwarded via tunnel between PE routers

     

    - last and I think best solution: treat MB routers as customers AS65500, you will run MPLS in core (PEs and Ps) which will take care of forwarding internal and also external traffic, you will create VRF instances on PE routers and BGP will distribute prefixes which you need. simple MPLS L3 VPN scenario

  • Amr M. el Adel-CCIE(RS,SP) 56 posts since
    Oct 3, 2011
    Currently Being Moderated
    6. Apr 22, 2013 2:13 AM (in response to Paul)
    Re: BGP configuration

    Hi,

     

    Paul, there is a blackhole on this network that the P routers don't know about the BGP routes coming from MB networks.

     

    So it is either you enable iBGP on these P routers (full mesh) or, run MPLS on the P and PE routers. No need for the VRF and creating MPLS L3 VPN, however it will work also.

     

    The scenario of enabling MPLS on core and PE is called MPLS tunnelling. In which, the MPLS core network (P routers) cares only about the reachability of the PEs' loopback interfaces and doesn't care about the destination network.

     

    Lab it and it will work.

Actions

More Like This

  • Retrieving data ...

Bookmarked By (0)