The preceding examples work with dial-on-demand routing as well as dial-backup. Set the metric on the backup interface to be less favorable than that on the primary. Also, set the administrative distance on the backup routing protocol to be greater than that of the primary. (Set both the metric and administrative distance to allow the idle timer to work.) The redistribution of a static route for the backed-up site is mandatory. Not only does it speed convergence, but it also is the controlling factor in directing traffic at the upstream dial-on-demand interface to trigger dialing. After the primary interface goes down, all knowledge of the site LAN is lost. If the backup server is originating routes for the site LANs at a much higher cost, these now come into effect.
Was this article helpful?