By default, route maps that are applied in router configuration mode using the neighbor route-map command are applied to only IPv4 unicast address prefixes. Route maps for other address families must be applied in address family configuration mode using the neighbor route-map command, as shown for the IPv6 address family. The route maps are applied either as the inbound or outbound routing policy for neighbors under the specified address family. Configuring separate route maps under each address family type simplifies managing complicated or different policies for each address family.

Router(config-router-af)# neighbor ipv6-address route-map map-name {in | out}

To configure a route map as it is needed when using a neighbour’s link local address in the BGP configuration the special command set ipv6 next-hop is used in the route map configuration.

Router(config) # route-map map-name [permit | deny] [sequence-number]
Router(config-route-map)# match ipv6 address prefix-list prefix-list-name
Router(config-route-map)# set ipv6 next-hop ipv6-address [link-local-address]

The last command overrides the next hop advertised to the peer for IPv6 packets that pass a match clause of a route map for policy routing.

The ipv6-address argument specifies the IPv6 global address of the next hop. It doesn’t need to be an adjacent router.

The link-local-address argument specifies the IPv6 link-local address of the next hop. It must be an adjacent router.

Note: The route map sets the IPv6 next-hop addresses (global and link-local) in BGP updates. If the route map is not configured, the next-hop address in the BGP updates defaults to the unspecified IPv6 address (::), which is rejected by the peer.

If you specify only the global IPv6 next-hop address (ipv6-address argument) with the set ipv6 next-hop command after specifying the neighbor interface (interface-type argument) with the neighbor update-source, the link-local address of the interface specified with the interface-type argument is included as the next-hop in the BGP updates. Therefore, only one route map that sets the global IPv6 next-hop address in BGP updates is required for multiple BGP peers that use link-local addresses.

As with IPv4 several BGP peers for which the same route maps and prefix lists apply can be configured as a peer group. This facilitates the configuration as well as optimizes the router’s handling of the corresponding rules.

By default, peer groups that are defined in router configuration mode using the neighbor peer-group command exchange only IPv4 unicast address prefixes. To exchange other address prefix types, such as IPv6 prefixes, you must activate peer groups using the neighbor activate command in address family configuration mode for the other prefix types, as shown for IPv6 prefixes.

Members of a peer group automatically inherit the address prefix configuration of the peer group.

IPv4 active neighbors cannot exist in the same peer group as active IPv6 neighbors. Create separate peer groups for IPv4 peers and IPv6 peers.

Peer groups in general are defined in global BGP configuration mode. The following command creates a multiprocotol BGP peer group:

Router(config)# router bgp autonomous-system-number
Router(config-router)# neighbor peer-group-name peer-group

Activating this peer group for IPv6 neighbours and adding neighbours to it must be done in the IPv6 address family configuration context. One can first add neighbors to a peer group and then later activate all neighbours at once by activating the corresponding peer group.

Router(config-router)# address family ipv6 [unicast]
Router(config-router-af)# neighbor ipv6-address peer-group peer-group-name
Router(config-router-af)# neighbor {ip-address \\
                                              | peer-group-name | ipv6-address} activate