WB 1 BGP 7.4 Multihop EBGP Peerings Query

Hello,

I am unable to understand why we are using same subnet IP on sw4 as lo1 /32 to configure ebgp peering with BB3 in AS 54?

and also 204.12.1.10 is not configured in BGP in BB3 as ebgp peering. 

please advice

 

Rack1R4#sh ip route 204.12.1.10

Routing entry for 204.12.1.0/24

  Known via "connected", distance 0, metric 0 (connected, via interface)

  Redistributing via eigrp 100

  Routing Descriptor Blocks:

  * directly connected, via FastEthernet0/0

      Route metric is 0, traffic share count is 1

 

Q

 

7.4 Multihop EBGP Peerings

• Create a new Loopback 1 interface on SW4 with the IP address

204.12.X.10/32, and advertise it into IGP.

• Configure an EBGP peering between SW4 and BB3 using this new

interface as the source of the peering.

Comments

  •  

    The purpose of the lab is to
    demonstrate the  function of the ebgp-multihop command. That is to
    alter the default TTL of 1 for EBGP Peers. To do this the peering has to be
    configured between two loopbacks. The loopbacks do not have to be on the same
    subnet as per the guide, but they can be. The important this is that there is
    ip connectivity from loopback to loopback.

     

    BB3 should be configured
    with for 204.12.1.10 as shown below.

     

    router
    bgp 54

    no
    synchronization

    bgp
    log-neighbor-changes

    neighbor
    RACKS peer-group

    neighbor
    RACKS remote-as 100

    neighbor
    RACKS ebgp-multihop 255

    neighbor
    2001:204:12:1::100 remote-as 100

    neighbor
    172.16.4.1 remote-as 54

    neighbor
    204.12.1.1 peer-group RACKS

    neighbor
    204.12.1.2 peer-group RACKS

    neighbor
    204.12.1.3 peer-group RACKS

    neighbor
    204.12.1.4 peer-group RACKS

    neighbor
    204.12.1.5 peer-group RACKS

    neighbor
    204.12.1.6 peer-group RACKS

    neighbor
    204.12.1.7 peer-group RACKS

    neighbor
    204.12.1.8 peer-group RACKS

    neighbor
    204.12.1.9 peer-group RACKS

    neighbor
    204.12.1.10 peer-group RACKS

    no
    auto-summary

     

     

     

  • JoeMJoeM ✭✭✭

    There are a couple of interesting things in this task.

     A  Router Trick:  

    • The loopback /32 is advertised internally by SW4 (longer prefix)
    • ...BB3 already knows to go to R4 via the directly connected  /24  route.
    • ...R4 knows the routes in both directions,  via /24 and /32.

    This config does us a favor.  By creating and advertising the /32 loopback, we do not having to advertised our internal addressing to BB3.    

    If we somehow gave BB3 the ip address for SW4's fastethernet interface, then we could use it for the eBGP connection.  But BB3 would then need to change its config also (pointing to the correct address). 

    We do not need loopbacks to make this happen.  But because we did, we do not have to mess with other routing advertisements between BB3 and the rest of the network.

     

     eBGP-multihop: 

    • for eBGP connections not directly connected
    • default is one hop.
    • We do not actually need as many hops as the solution shows.

    In this case, we have four (4) hops to get from router to router.

    Try playing with the hops on this.   In my testing, failure happens at 3 hops.   ebgp-multihop 3

     

  • The purpose of the lab is to
    demonstrate the  function of the ebgp-multihop command. That is to
    alter the default TTL of 1 for EBGP Peers. To do this the peering has to be
    configured between two loopbacks. The loopbacks do not have to be on the same
    subnet as per the guide, but they can be. The important this is that there is
    ip connectivity from loopback to loopback.

    Hi,

    Yes, we can modify the TTL value up to 255 when configuring eBGP peers that are not in the common segment. Without altering TTL they don't peer with each others. You can give any value that is greater than your required value. For the TCP connectivity between these peers, you can have any kind of IGP or static routing in place. 

    HTH

  • Dear My Friends,

    Thank you very much for your kind reply but still i am finding reason to configure same subnet IP with /32 loopback 1 on SW4 where /24 subnet already configured between R4 and BB3.

    so why we need to configure same ip on SW4? how it will ping or reachanble till BB3 in order to configure ebgp between SW4 and BB3?

     

    thanks

    Sourabh

  • JoeMJoeM ✭✭✭

    Dear My Friends,

    Thank you very much for your kind reply but still i am finding reason to configure same subnet IP with /32 loopback 1 on SW4 where /24 subnet already configured between R4 and BB3.

    so why we need to configure same ip on SW4? how it will ping or reachanble till BB3 in order to configure ebgp between SW4 and BB3?

     

    thanks

    Sourabh

    Hi Sourabhd,

    I give the reason in my post above.  It is a router trick, that has the benefit of keeping the loopback disguised. No internal routing needs to be given to BB3. 

    With this method, BB3 does not need to see the /32 (but it does see the /24).   Think of it like a default-route for that subnet.  R4 then passes it on to the longer prefix.

     

    ROUTING STEPS:

    1.  BB3 knows about its directly connected subnet 204.12.1.0/24 

    2.  SW4 also sees the /24, and uses it to reach BB3

    3.  R4 (and the rest of the internal network) see the /24 and the longer prefix  /32

     

    CONNECTIVITY TEST:

    Do a couple of pings from SW4 to BB3.  We must have this connectivity, for the eBGP neighbors to come up.

    First, a normal ping should fail, because BB3 does not know how to reach SW4 (155.x.108.10)

                   ping 204.12.1.254

    Now try sourcing the ping from the new /32 loopback (success)

                  ping 204.12.1.254  source lo1

Sign In or Register to comment.