3.6.c Implement and troubleshoot OSPFv3 address-family support

  • 3.6.c [i] IPv4 address-family
  • 3.6.c [ii] IPv6 address-family

The OSPFv3 address families feature enables both IPv4 and IPv6 unicast traffic to be supported . With this feature, users may have two router processes per interface, but only one process per AF. If the IPv4 AF is used, an IPv4 address must first be configured on the interface, but IPv6 must be enabled on the interface. A single IPv4 or IPv6 OSPFv3 process running multiple instances on the same interface is not supported.

Users with an IPv6 network that uses OSPFv3 as its IGP may want to use the same IGP to help carry and install IPv4 routes. All routers on this network have an IPv6 forwarding stack. Some (or all) of the links on this network may be allowed to do IPv4 forwarding and be configured with IPv4 addresses. Pockets of IPv4-only routers exist around the edges running an IPv4 static or dynamic routing protocol. In this scenario, users need the ability to forward IPv4 traffic between these pockets without tunneling overhead, which means that any IPv4 transit router has both IPv4 and IPv6 forwarding stacks (e.g., is dual stack). This feature allows a separate (possibly incongruent) topology to be constructed for the IPv4 AF. It installs IPv4 routes in IPv4 RIB, and then the forwarding occurs natively. The OSPFv3 process fully supports an IPv4 AF topology and can redistribute routes from and into any other IPv4 routing protocol.

An OSPFv3 process can be configured to be either IPv4 or IPv6. The address-family command is used to determine which AF will run in the OSPFv3 process, and only one address family can be configured per instance however multiple instances of OSPFv3 can be enabled on single interface . Once the AF is selected, users can enable multiple instances on a link and enable address-family-specific commands.

Different instance ID ranges are used for each AF (it is carried inside packet header as instance ID field and that’s unique to IPv6). Each AF establishes different adjacencies, has a different link state database, and computes a different shortest path tree. The AF then installs the routes in AF-specific RIB. LSAs that carry IPv6 unicast prefixes are used without any modification in different instances to carry each AFs’ prefixes. The IPv4 subnets configured on OSPFv3-enabled interfaces are advertised through intra-area prefix LSAs, just as any IPv6 prefixes. External LSAs are used to advertise IPv4 routes redistributed from any IPv4 routing protocol, including connected and static. The IPv4 OSPFv3 process runs the SPF calculations and finds the shortest path to those IPv4 destinations. These computed routes are then inserted in the IPv4 RIB (computed routes are inserted into an IPv6 RIB for an IPv6 AF).

Because the IPv4 OSPFv3 process allocates a unique pdbindex in the IPv4 RIB, all other IPv4 routing protocols can redistribute routes from it. The parse chain for all protocols is same, so the ospfv3 keyword added to the list of IPv4 routing protocols causes OSPFv3 to appear in the redistribute command from any IPv4 routing protocol. With the ospfv3 keyword, IPv4 OSPFv3 routes can be redistributed into any other IPv4 routing protocol as defined in the redistribute ospfv3 command.

In case of using OSPF for IP VPNs as PE-CE protocol , the down bit helps prevent routing loops between MP -BGP and OSPF, but not when external routes are announced, such as when redistribution between multiple OSPF domains or when external routes are injected in an area that is dual-homed to the provider network. The PE router redistributes an OSPF route from a different OSPF domain into an OSPF domain as an external route. The down bit is not set because LSA Type 5 does not support the down bit. The redistributed route is propagated across the OSPF domain.

Adam, Paul (2014-07-12). All-in-One CCIE V5 Written Exam Guide (Kindle Locations 3492-3502).  . Kindle Edition.