Deployment Guide

Table Of Contents
Establishing Sessions with BGP Neighbors
Before configuring BFD for BGP, you must first configure BGP on the routers that you want to interconnect.
For more information, refer to Border Gateway Protocol IPv4 (BGPv4).
For example, the following illustration shows a sample BFD configuration on Router 1 and Router 2 that use eBGP in a transit
network to interconnect AS1 and AS2. The eBGP routers exchange information with each other as well as with iBGP routers to
maintain connectivity and accessibility within each autonomous system.
Figure 16. Establishing Sessions with BGP Neighbors
The sample configuration shows alternative ways to establish a BFD session with a BGP neighbor:
By establishing BFD sessions with all neighbors discovered by BGP (the bfd all-neighbors command).
By establishing a BFD session with a specified BGP neighbor (the neighbor {ip-address | peer-group-name}
bfd command)
BFD packets originating from a router are assigned to the highest priority egress queue to minimize transmission delays.
Incoming BFD control packets received from the BGP neighbor are assigned to the highest priority queue within the control
plane policing (COPP) framework to avoid BFD packets drops due to queue congestion.
BFD notifies BGP of any failure conditions that it detects on the link. Recovery actions are initiated by BGP.
BFD for BGP is supported only on directly-connected BGP neighbors and only in BGP IPv4 networks. Up to 128 simultaneous
BFD sessions are supported
As long as each BFD for BGP neighbor receives a BFD control packet within the configured BFD interval for failure detection,
the BFD session remains up and BGP maintains its adjacencies. If a BFD for BGP neighbor does not receive a control packet
within the detection interval, the router informs any clients of the BFD session (other routing protocols) about the failure. It
then depends on the individual routing protocols that uses the BGP link to determine the appropriate response to the failure
condition. The typical response is to terminate the peering session for the routing protocol and reconverge by bypassing the
failed neighboring router. A log message is generated whenever BFD detects a failure condition.
1. Enable BFD globally.
CONFIGURATION mode
bfd enable
Bidirectional Forwarding Detection (BFD)
155