IPv6 Addressing Requirements for a Dual-Stack Network
You need to implement two types of addressing for IPv6 in a subscriber access network:
WAN link addressing—For the WAN interface on the customer premises equipment (CPE) (CPE upstream interface).
Subscriber LAN addressing—For devices connected to the CPE on the subscriber LAN (CPE downstream interfaces).
Figure 1 shows where WAN link addressing and subscriber addressing are assigned in a dual-stack network.
You can use the following methods for assigning IPv6 addresses:
For WAN link addressing, you can use Neighbor Discovery Router Advertisement (NDRA) or Dynamic Host Configuration Protocol for IPv6 (DHCPv6) identity association for nontemporary addresses (IA_NA) to provision a global IPv6 address.
For subscriber LAN addressing, you can use DHCPv6 prefix delegation to provision global IPv6 addresses to subscribers on the LAN.
Alternatives to Using a Global IPv6 Address on the CPE WAN Link
If the CPE is supplied by or recommended by the service provider, you do not need to provision a unique global IPv6 address on the CPE. In this case, the broadband network gateway (BNG) can use the loopback interface to manage the CPE. You can use one of the following methods to provision an address on the loopback interface:
Link-local IPv6 address—Can be used on Point-to-Point Protocol over Ethernet (PPPoE) access networks. The link-local address is provisioned by appending the interface identifier negotiated by the Internet Protocol version 6 Control Protocol (IPv6CP) with the IPv6 link-local prefix (FE80::/10).
Address derived from DHCPv6 prefix delegation—Can be used on PPPoE access networks or on DHCP access networks. If you use DHCPv6 prefix delegation for subscriber addressing, the CPE can use the prefix it receives from the BNG to assign an IPv6 address on the loopback interface between the CPE and the BNG. This address can be used to manage the CPE, and the CPE uses it as a source address when it communicates with the BNG.