Users Guide

1426 BGP
IPv6 prefixes can be originated through route redistribution or a network
command. Both can be configured with a route map to set path attributes.
BGP can also originate an IPv6 default route. Default-origination can be
neighbor-specific. IPv6 routes can be filtered using prefix lists, route maps
with community lists, and using AS path access lists. BGP can compute IPv6
routes with up to 16 ECMP next hops.
IPv6 Peering Using A Link Local Address
Link local addresses are one class of IPv6 address that can be used as a BGP
peer address. Allowing link local addresses to be used as peer addresses
introduces some complications. These are discussed here:
First, consider whether it even makes sense to use a link local address as a peer
address. As its name implies, a link local address has link scope; it is only valid
on a single link. This characteristic implies that a BGP peer identified by a
link local address must be attached to a local link (i.e., shares a Layer-2
broadcast domain with the router where the peer is configured). This
restriction is typically met for external peers (with the exception of external
peers configured with ebgp-multihop). Internal peers are typically not on a
local link. Even when two BGP speakers share a common link, a loopback
address is often assigned as the peer address to avoid tying the fate of the
adjacency to a single link or interface. However, it is possible to configure
internal peers using an IPv6 address configured on the link that connects
them.
Another feature of link local addresses that makes them less than ideal for
BGP peering is that they are auto-generated, typically from the local MAC
address. Because each BGP peer is configured with a specific IP address, the
peer address must be known at configuration time. This is in contrast to
OSPFv3 adjacencies, where neighbor addresses are learned dynamically. If the
peer's MAC address changes (for example, if a router fails and is replaced with
new hardware), the link local address may change and the BGP configuration
will need to be updated. Dell EMC Networking allows discovery of BGP peers
using link-local addresses through the configuration of listen ranges.
To use link local addresses for eBGP peers, an administrator must use the
next-hop-self option. Normally, when a BGP speaker forwards an external
route to internal peers, it retains the BGP NEXT_HOP. If the NEXT_HOP is
a link local address, internal peers will be unable to resolve it and thus not be
able to use these routes. The router must therefore be configured to change