Routing

 View Only
last person joined: 2 days ago 

Ask questions and share experiences about ACX Series, CTP Series, MX Series, PTX Series, SSR Series, JRR Series, and all things routing, including portfolios and protocols.
  • 1.  Adding 1 multihop peer to EGBP

    Posted 12-21-2020 11:53

    Hi, I have been reading through the documentation at BGP Multihop Sessions - TechLibrary - Juniper Networks and trying to understand how this best relates to my needs.

    At present I have a simple EGBP configuration that does a failover default route between two ISP's and brings in routes from INET 2. All go through directly connected peers and I haven't had a need to do a loopback address. We export 1 class C range with our AS. I keep this as simple as possible because we don't have any Juniper experts. 

    But we needed to add in Comcast Xfinity streaming routes that require a multi hop peer along with a normal neighbor and I am not sure of the best way to do this.

    Key questions:

    1. The documentation is using different logical sessions which I don't think I need. Is this correct? These routes should be additions to our routing table but not a separate process.

    2. Can I add a loopback for the multihop address without needing to do the same for all peers?

    3. Can I configure this under the current external peers group or do I need to add a new group for multihop specific configurations?

    4. Any other pitfalls and issues I should be aware of?

    This is bgp parts of the config as is before changing anything.

    Comcast gave me a cisco example. 

    !Physical CPE port

    Interface gige x/x

       Description Connection to Comcast CPE

        No ip address

        No shut

     

    !Logical interface used to connect with CPE

    interface gige x/x.<vlan-id>

       description Connection to Comcast network

       encap dot1Q <vlan-id>

       ip address <other side of /30> 255.255.255.252

       mtu 1500

      < Add other interface parameters as needed>

     

    !Static route needed for FULL-ROUTE customer to build MH-BGP session with Ibone PE ***

    ip route <ip address of PE loopback1(Comcast to provide)> 255.255.255.255 <interface.<vlan-id>> <SUR interface ip address(Comcast to provide)>

     

    !Used to aggregate prefixes to advertise via BGP

    ip route <ip address agg> /<mask> null0

     

    ! Used to limit routes announced to SUR

    ip prefix-list PERMIT-TO-COMCAST seq 5 permit <agg to announce>

     

    !Used to NOT announce anything to the 2nd BGP session.

    ip prefix-list DENY-TO-COMCAST seq 5 deny 0.0.0.0/0

     

    !Primary bgp session with SUR

    router bgp <customer ASN>

      no synchronization

      bgp router-id <loopback0>

      network <ip address agg/mask>

     

      neighbor <SUR interface ip address> description Peering Session with Comcast

      neighbor <SUR interface ip address> remote-as 7922

      neighbor <SUR interface ip address> send-community

      neighbor <SUR interface ip address> prefix-list PERMIT-TO-COMCAST out

     

    !Second FULL-ROUTE customer session

      neighbor <PE Loopback ip address> description Peering Session with Comcast

      neighbor <PE Loopback ip address> remote-as 7922

      neighbor <PE Loopback ip address> ebgp-multihop 255

      neighbor <PE Loopback ip address> send-community

      neighbor <PE Loopback ip address> prefix-list DENY-TO-COMCAST o

    routing-options {
    static {
    route 172.16.1.0/24 {
    next-hop 172.17.3.1;
    retain;
    no-readvertise;
    }
    }
    autonomous-system 62736;
    }
    protocols {
    bgp { 

    group external-peers {

    type external;
    export Landmark_Export;
    neighbor 132.198.255.210 {
    description V-Tel_Inet-2;
    peer-as 1351;
    }
    neighbor 216.66.110.77 {
    description V-Tel_Internet;
    peer-as 17356;
    }
    neighbor 4.53.87.245 {
    description Century_Link;
    peer-as 3356;
    }

    policy-options {
    prefix-list Landmark_BGP_Export {
    198.54.211.0/24;
    }
    policy-statement Landmark_Export {
    term advertise_Landmark {
    from {
    prefix-list Landmark_BGP_Export;
    }
    then accept;
    }
    then reject;
    }
    }



  • 2.  RE: Adding 1 multihop peer to EGBP

    Posted 12-22-2020 15:07

    if you are simply needing to bring up another bgp session, then you can probably add it to the existing bgp group.  i would just add import and export policies to it to only allow/disallow what you do or do not want to advertise.  then add the required amount of TTL hops for it since you mentioned it is bgp mhop.  looks like you have 255, but you probably don't need it that high.



    ------------------------------
    Aaron Gould
    Senior Network Engineer
    aaron@gvtc.com
    https://www.linkedin.com/in/agould123/
    ------------------------------



  • 3.  RE: Adding 1 multihop peer to EGBP

    Posted 12-29-2020 12:29

    1) No, you don't need to create logical systems. I don't know why they are using logical systems in the document. 

    2) yes, you can add a session loopback to loopback with multihop, without having to make any changes to other peers. 

    3) No, you don't need to have a different BGP group.  You might want to have a separate group  just for management purposes, but it is NOT required. 

    For example, if you create a second group, you can do show bgp summary group <multihop_group_name> and show bgp summary group <singlehop_group_name> to get separate information for either group VS. doing  show bgp summary and having all neighbor listed together.  You could also have group policies that are different for the multihop and the single hops neighbors (if you later add more to each group).  

    4) I tested it like shown below to see if I could think of anything, but other than maybe watch out for what you are advertising and accepting on each session (policies), and which route is preferred, so not create sub-optimal routing  (if you have multiple physical connections -obviously not the case in my test setup) , I don't think you need to worry about anything. 

    Regards,