Using Multiple IPv6 WAN Connections

With pfSense® software version 2.1 and later Multi-WAN with IPv6 is possible provided that multiple ISPs or tunnels are setup and working. See Using IPv6 with a Tunnel Broker if help is needed setting up a tunnel.

Throughout this document “Second WAN” refers to the second or additional interface with IPv6 connectivity. It might be the “real” interface if native connectivity is present, or a tunnel if a tunnel broker is being used. Keep that in mind, as it will make understanding the rest of this document easier.

IPv6 Multi-WAN Caveats

Traditionally with IPv6 everything is routed and there is no NAT. That’s great for connectivity, and for businesses or locations that can afford PI space and a BGP peering. It doesn’t work so well in practice for home users.

Network Prefix Translation (NPt) will allow one subnet to be used for LAN and have full connectivity with that subnet via its “native” WAN, and also have it translated on the additional WANs so it appears to originate there. While not “true” connectivity for the LAN subnet via that path, it is better than no connectivity at all if the primary WAN is down.

This may not work at all for completely dynamic IPv6 types where the subnet is not static. (DHCP-PD, etc)

Requirements

To setup Multi-WAN for IPv6 the following items are needed:

  • Two WANs, and IPv6 connectivity setup on both.

  • Gateways added to System > Routing for both, and confirmed connectivity on both.

  • LAN using a static routed /64 or similar

  • A routed /64 or larger block available on both WANs

Setup

  • Under System > Routing on the Gateway Groups tab, add Gateway Groups for the V6 gateways, this works like IPv4 Multi-WAN

  • Under System > General, ensure there is an IPv6 DNS server set for each IPv6 WAN. Again, the same as IPv4

  • Add an NPt entry under Firewall > NAT on the NPt tab:

    • Interface: Secondary WAN (or tunnel if using a broker)

    • Internal IPv6 Prefix: The LAN IPv6 subnet

    • Destination IPv6 Prefix: The second WAN’s routed IPv6 subnet (not the /64 of the WAN interface itself – the /64 routed to pfSense on that WAN by the upstream)

    What this does is act similar to 1:1 NAT for IPv4. As traffic leaves the second WAN, if it’s coming from the LAN subnet, it will be translated to the equivalent IP in the other subnet. For example if 2001:xxx:yyy::/64 is on LAN, and 2001:aaa:bbb::/64 is the routed subnet on the second WAN, then 2001:xxx:yyy::5 would appear as 2001:aaa:bbb::5 if the traffic goes out the second WAN.

  • As with IPv4 the Gateway Groups must be used on the LAN firewall rules. Edit the LAN rules for IPv6 traffic and make them use the gateway group, making sure to have rules for directly connected subnets/VPNs without a gateway set so they are not policy routed. More information on that is on the Multi-WAN page.

Alternate Tactics

Some may prefer to use a “private” IPv6 subnet in LAN from the fc00::/7 space, and setup NPt for both WANs.