In both IPv4 and IPv6 networks, you can configure multiple routes to a common destination and build a redundant (and resilient) network topology by assigning preference values to routes. To do this, set the preference value of one router to a higher preference level (using a lower value, for example 1 indicates higher preference than 2), and thus create an alternate backup route should the route of lower preference value (the preferred router) become disabled.

Defining multiple routes with equal preference values allows the potential for balancing individual network dialogs across each of the next-hops. For example, if there are two equal preference next-hops within the same network prefix, a balancing algorithm is employed such that the first dialog would use Router-1 and the second dialog would use Router-2. As with alternate routes, in the event that one of these parallel routes experiences a failure, all dialogs can be re-assigned to the other route.

Multiple "assigned" default routes can also be configured and/or assigned to a specific network prefix. Default routes that are configured without a route mask are considered "system" default routes and will be selected in the absence of an assigned default route (one where the route mask has been defined).

Note that redundant configurations, such as those described here, do not always imply the capability for dialog resiliency. Redundancy and resiliency are not synonymous; nor does one imply the other. Achieving seamless dialog recovery (resiliency) in the event of a route failure requires that the networks within the topology eliminate every single point of failure. Accomplishing this requires a redundant topology that provides multiple routes to desired remote destinations from within the same network prefixes.

The following subsections show sample topologies you can configure with MCP TCP/IP that support multiple routes to a destination.