BGP address-family

Hi,

i am trying to understand about bgp address-family but not understand, by default bgp form neighbor with TCP pot 179,also with default config of verion 4.but i dont know bgp use unicast or multicast. many webiste says that bgp address family is form neighbor with particular address family.

Please any one explain me simply about this uses

Comments

  • Hi,

    BGP uses TCP port 179 and unicast to establish peer connection. This is general.

    Address family is used to exchange different families, like IPv6,VPN4 etc with the peer. You can activate different peers in different address families. Ipv4 is on by default but it is considered good practice to use address family for ipv4 as well.

    It adds more modularity to the configuration and easy to read. All the new IOS versiont support it.

    There are alot of docs regarding this on cisco website and may be here on the forum as well.

    HTH

  • peetypeety ✭✭✭

    BGP has more separation between neighbor establishment and route advertisement/propagation than the IGPs.  In this case, you're correct that BGP neighbors are established with TCP/179 and unicast packets.  Once the neighbors are established, then routes can be exchanged, and these routes can be of various families or "flavors": ipv4 unicast, ipv4 multicast, ipv6 unicast, vpnv4, etc.  The typical usage of BGP is ipv4 unicast, and that's perhaps the "default" address family when you first activate BGP, but you definitely have the opportunity to choose which families you activate.

  • >>>but i dont know bgp use unicast or multicast.

    Feroz,

    I wanted to point out something that is usually a point of confusion.  While BGP can exchange multicast IPv4 address family (AF), it is never used for actually forwarding multicast traffic or maitaining multicast state abour source or interested recievers.  (BGP is not a multicast routing protocol in the traditional sense)  The BGP multicast AF is used to influence the RPF validation for multicast traffic that arrives at the local router.  EDIT: there are cases (such as MVPN) where BGP can exchange multicast state information used other AF's in the context of MPLS; see post below.

    The outgoing interface list (OIL) where mutlicast is actually forwarded is determined by some flavor of PIM. (via PIM join from the downstream router (hosts use IGMP to its local router) that has signaled their desire to recieve the particular group traffic.)

    In the Cisco world BGP peers exchange IPv4 unicast AF by default but this can be disabled.  Any other AF will need to be activated on a per neighbor basis. 

    Take care,

    Charles

     

     

     

  • Good post, cstizza.

     

    With NG-MVPN, BGP can be used to assign flows to MP-LSP so it is more involved in the multicast forwarding. It can even replace PIM in some cases.

  • Hi Daniel,

    Thanks for pointing that out..I was not thinking in the context of an Multicast VPN/MPLS but glad you mentioned it.

    To the best of my knowledge Cisco's implementation has a specific address family called "ipv4 mvpn" and does not use "ipv4 multicast" af for this state informaton.   (But to your point it is case where BGP is in fact being using to convey multicast state.)

    Take care,

    Charles

Sign In or Register to comment.