Information Technology Reference
In-Depth Information
The loopback address is 0:0:0:0:0:0:0:1, or ::1 in abbreviated form. A node may use this
address to send packets to itself. This address should be considered to have link-local scope
over a virtual link from the node back to itself. Packets should never be sent by a node with
either the source or destination addresses set to the loopback address. An IPv6 router should
not forward packets that contain the loopback address as either the source or destination
address.
MP-BGP Extensions for IPv6 NLRI
The deployment of IPv6 in a global network requires an Exterior Gateway Protocol (EGP)
to provide full global reachability. The decision was to include support for IPv6 NLRI and
to allow IPv6 transport for MP-BGP. This allowed the reuse of operational experience
gained through years of work with BGP. The BGP extension to carry various types of
routing information is called multiprotocol BGP, or MP-BGP. This functionality is
discussed in Chapter 2, “Understanding BGP Building Blocks,” in the section “BGP
Capabilities.”
The difference between BGPv4 and MP-BGP is the ability to carry prefix information for
multiple protocols, such as IP multicast, CLNS, MPLS labels, IPv4 VPN, and IPv6. This
information is advertised using address families. Prefix information is advertised using the
BGP attribute MP_REACH_NLRI and is withdrawn using the MP_UNREACH_NLRI
attribute.
The Address Family Identifier (AFI) for IPv6 is 2, and the Subsequent Address Family
Identifiers (SAFIs) are 1 for unicast, 2 for multicast, and 3 for both unicast and multicast.
The next two sections discuss the concept of a dual-stack deployment with IPv4 and IPv6
running concurrently and the protocol impact of integrating IPv6 support into MP-BGP.
Dual-Stack Deployment
The term
dual stack
is frequently used when discussing IPv6 deployment. Dual stack refers
to running IPv4 and IPv6 concurrently in the router or device. This is commonly considered
the preferred method for migrating from IPv4 to IPv6.
The use of a dual-stack deployment places additional load on the routers in the network.
There are increased resource requirements to maintain both the IPv4 and IPv6 routing
tables. This includes separate RIBs and FIBs. Most likely, you will also need multiple IGPs.
In the case of BGP, there are increased memory requirements to maintain the additional
prefix information. There is also additional processing to maintain more BGP peering
sessions. You should configure separate BGP sessions for IPv4 and IPv6 prefix information.
This ensures next-hop reachability for each address family.
