cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1477
Views
0
Helpful
6
Replies

VPN Route-Reflector

Hi All,

         I have a number of PEs peering with a RR, the aim is for these RR-clients to exchage their customers routes via the RR.

Now looking at the configuration below, i cannot see the need as to  why "address-family ipv4" should be present in the config.

Can someone please tell me why if at all it should be in place?. (This is just one of the clients).

router bgp XX

bgp router-id 2.2.2.2

no bgp default ipv4-unicast

bgp log-neighbor-changes

neighbor 172.16.123.3 remote-as 9

neighbor 172.16.123.3 update-source Loopback0

!

address-family ipv4

  neighbor 172.16.123.3 activate

  neighbor 172.16.123.3 send-community

  neighbor 172.16.123.3 route-reflector-client

  exit-address-family

!

address-family vpnv4

  neighbor 172.16.123.3 activate

  neighbor 172.16.123.3 send-community

neighbor 172.16.123.3 route-reflector-client

exit-address-family

BR,

O.A

6 REPLIES 6
blau grana
Rising star

Hello,

If you want propagate only vpnv4 prefixes, IPv4 AF is not necessary in your config.

Best Regards

Please rate all helpful posts and close solved questions

Best Regards Please rate all helpful posts and close solved questions

Hi Blau,

            Many thanks for your reply to my query. Unfortunately my question wasn't answered yet.

Dividing the configs into three, i know the first part establishes the TCP session, the third part enables MP-BGP between the PEs.

Now the second part is there "address-family ipv4", with or with it routes are reaching the right places, when do we really need it then?.

I will appreciate a more elaborate answer as why/when/how do we need it?.

Thanks,

O.A

Hello,

Sorry I did not realize that you wanted also explanation

BGP use address family model, so multiple AF were introduced. Basically it means what AF is used, that prefixes are exchanged:

IPv4 AF - used when IPv4 prefixes are exchanged

IPv6 AF - used when IPv6 prefixes are exchanged

VPNv4 AF - used when VPNv4 prefixes are exchanged

VPNv6 AF - used when VPNv6 prefixes are exchanged

For more details read thi post from Peter Paluch, which is extremly helpful:

https://supportforums.cisco.com/message/3615165#3615165

And also cisco documentation:

http://www.cisco.com/en/US/docs/ios/12_4t/ip_route/configuration/guide/tbrbover.html#wp1049730

Best Regards

Please rate all helpful posts and close solved questions

Best Regards Please rate all helpful posts and close solved questions

hi fellow

vpnv4 AF, as mentionned earlier will propagate vpnv4 routes on an MPLS backbone, by adding 1 bottom stack mpls label to identify destination vpn when packet is reaching outbound PE.

ipv4 AF does not use any labels. just a classical BGP peerings exchanging routes.

vpnv4 usually used on mpls bgp free core by ISP to isolate traffics between customers.

with ipv4 AF, all routers on the path have to be full mesh ( or connected to RR), and must be aware of all the routes.

so you could use one or another or both, depending on your setup.

usually on a BGP free core, vpnv4 is all that you need. now your customer might not need ipv4 AF. you have to understand exactly the network design for that, or give us a bit more info about your topology.

HTH

medhat elsaeed
Beginner

Hi,

you will need the AF IPV4 if your PE routers are exchanging routes in the Global routing table, for example some ISPs are ruunning their internet customers in the Global routing table ( which means that the customer interface in the Global not under VRF )

the VPNV4 AF will exchange the VRF routes not the Global one

Regards.

Martin12
Beginner

quickly to the point : 'addess-family ipv4' = af ipv4 [vrf NOTHING/NULL/GLOBAL]
with MPBGP you have VPNv4 peers where you would send ext community and exchange NLRIs(prefixes) but what if there is one peer that does not support / is not configured for VPNv4 capability ?
You will need to configure the BGP peer with classic IPv4 (vrf global) or in IPv4 VRF X for VRF lite (where the VRF is only local, no RDs, no RTs, you can match VRF A on one side and VRF B on another, the router won't know ...) -- so then it depends in which VRF routing table you find your BGP peer to send him TCP and maybe it is the global one -- then you will need that default a-f IPv4 and no bgp default ipv4-unicast is good in case that you are sure you don't need that a-f ;)
Content for Community-Ad