The symmetric or asymmetric question is linked to the configuration of redundancy for the routers serving this cluster. If the single subnet model is used, the router configuration in the direction of the cluster also follows the asymmetric model. Virtual Router Redundancy (VRRP) is configured with one router as the primary and the other router as the secondary. If multiple subnets are configured, it is common practice to make one router the primary for some of the subnets and the other router as the primary for the rest. Note that VRRP should be configured with a failover latency greater than the latency required for the Layer 2 loop avoidance protocol to prevent LAN failures from disturbing the wider network. Typical defaults are between two and three seconds, which should be enough to prevent LAN failures in a simple well configured spanning tree.

The cluster subnet is exported to OSPF through the interfaces to the core so that the devices are reachable, but OSPF needs to know which router interface to use for the packets directed to the cluster. For proper operation, the link to the primary router must be the preferred OSPF path. If the primary router fails but the link to the core is still operational, packets do not reach the cluster until the neighbor adjacency times out. Making these timeouts too small makes the protocol overly sensitive and may still provide inadequate results.

The probability of a VRRP interchange that occurs asymmetrically is arguably lower than a router failure that leaves the physical link state unchanged. Some implementations address this by allowing the link state of different interfaces to be coupled. These techniques are also applicable to the OSPF solution but are typically proprietary. Combining the link state of different interfaces with the decoupling route core disruption from local failure are arguments for this configuration.