Unicast Routing Protocols
Unicast routing protocols treat each IP network as an interface. The interface corresponding to the primary subnet is the active interface, and the interfaces corresponding to the secondary subnet are passive subnets.
For example, in the case of OSPF, the system treats each network as an interface, and hello messages are not sent out or received over the non-primary interface. In this way, the router link state advertisement (LSA) includes information to advertise that the primary network is a transit network and the secondary networks are stub networks, thereby preventing any traffic from being routed from a source in the secondary network.
Interface-based routing protocols (for example, OSPF) can be configured on per VLAN basis. A routing protocol cannot be configured on an individual primary or secondary interface. Configuring a protocol parameter on a VLAN automatically configures the parameter on all its associated primary and secondary interfaces. The same logic applies to configuring IP forwarding, for example, on a VLAN.
Routing protocols in the multinetted environment advertise the secondary subnets to their peers in their protocol exchange process. For example, for OSPF the secondary subnets are advertised as stub networks in router LSAs. RIP also advertises secondary subnets to its peers residing on the primary subnet.
This section describes the behavior of OSPF in an IPv4 multinetting environment:
- Each network is treated as an interface, and hello messages are not sent out or received over the non-primary interface. In this way, the router LSA includes information to advertise that the primary network is a transit network and the secondary networks are stub networks, thereby preventing any traffic from being routed from a source in the secondary network.
- Any inbound OSPF control packets from secondary interfaces are dropped.
- Direct routes corresponding to secondary interfaces can be exported into the OSPF domain (by enabling export of direct routes), if OSPF is not enabled on the container VLAN.
- When you create an OSPF area address range for aggregation, you must consider the secondary subnet addresses for any conflicts. That is, any secondary interface with the exact subnet address as the range cannot be in another area.
- The automatic selection algorithm for the OSPF router ID considers the secondary interface addresses also. The numerically highest interface address is selected as the OSPF router-id.
This section describes the behavior of the Routing Information Protocol (RIP) in an IP multinetting environment:
- RIP does not send any routing information update on the secondary interfaces. However, RIP does advertise networks corresponding to secondary interfaces in its routing information packet to the primary interface.
- Any inbound RIP control packets from secondary interfaces are dropped.
- Direct routes corresponding to secondary interfaces can be exported into the RIP domain (by enabling export of direct routes), if RIP is not enabled on the container VLAN.
There are no behavioral changes in the BGP in an IP multinetting environment.
This section describes a set of recommendations for using BGP with IP multinetting:
- Be careful of creating a BGP neighbor session with a BGP speaker residing in secondary subnet. This situation can lead to routing loops.
- All secondary subnets are like stub networks, so you must configure BGP in such a way that the BGP next hop becomes reachable using the primary subnet of a VLAN.
- When setting the BGP next hop using an inbound or outbound policy, ensure that the next hop is reachable from the primary interface.
- A BGP static network's reachability can also be resolved from the secondary subnet.
- Secondary interface addresses can be used as the source interface for a BGP neighbor.
- Direct routes corresponding to secondary interfaces can be exported into the BGP domain (by enabling export of direct routes).
This section describes the behavior of IS-IS in an IPv4 multinetting environment:
- IS-IS includes all the interface addresses in its reachability information. Adjacency is established only based on the primary interface address. If the adjacency-check option is disabled by the disable isis adjacency-check command, then IS-IS adjacency is established irrespective of the subnet address match.