Notice that Router A has learned about 0.0.0.0 as an external route with metric of 10. The gateway of last resort is set to 220.127.116.11 as expected. Thus, its default route is the E0 interface of Router C that has a default route in Router E.
Designing OSPF On-Demand Circuits
On-demand circuits can come in many different forms, from ISDN to SVCs. They tend to be implemented in one of two ways. First, they are put in place as a backup for the dedicated circuit, or second, they are for sites that require connectivity, just not all the time.
The OSPF on-demand circuits is an enhancement to the original OSPF protocol that allows efficient operation over on-demand circuits like ISDN, X.25 SVCs, and dial-up lines. This feature was first introduced in RFC 1793, Extending OSPF to Support Demand Circuits. It is fully supported by Cisco in all releases of their IOS.
This feature is useful when you want to connect telecommuters or branch offices to an OSPF backbone at a central site. As the pricing of demand circuits has gone down and the criticality of networks has increased, many network designers are turning to demand circuits as a means of back up.
Prior to this feature, OSPF periodic hello and LSAs updates would be exchanged between routers that connected via the on-demand link, even when no changes occurred in the hello or LSA information. This is, of course, normal operation for the OSPF protocol, but it has the unwanted side effect of causing the demand circuit to remain active because there was always interesting traffic to route across it.
However, with this new RFC, periodic hellos are suppressed and the periodic refreshes of LSAs are not flooded over the demand circuit. These packets bring up the link only when they are exchanged for the first time, or when a critical change occurs in the information they contain.
This suppression allows the demand circuit to be released. This is extremely important because most of the demand circuits have usage fees relating to them in either the length of use or amount of use and sometimes even both.
In this case, OSPF for on-demand circuits allows the benefits of OSPF over the entire domain, without excess connection costs. Periodic refreshes of hello updates, LSA updates, and other protocol overhead traffic is prevented from enabling the on-demand circuit when there is no real data to transmit.
Golden Rules for Designing Demand Circuits
As with every configuration of OSPF, there are a series of golden rules that you must be aware of before proceeding on. This is a list of them for demand circuits:
Dial On-Demand Design Scenarios
There are a number of common scenarios that will be encountered if you are planning on using this feature of OSPF. The first two are ways (NOT the best ways!) to implement OSPF. In the following scenarios, Site Router A is the router equipped with on-demand dialing.
Design Scenario #1: Site Router Is in Two Areas (Neither Is Area 0)
This approach does not work as the LAN interface cannot be in more than one area, as shown in Figure 6-10. There is no exchange of link-state information between areas 1 and 2.
As shown in Figure 6-10, the site router is located in two different OSPF areas with neither of them being area 0. However, if the site LAN is not included in the OSPF routing, and its routing information is injected with a static route either at the site router or at the distribution router, this could be made to work, although it is not the most optimal OSPF network design.
Design Scenario #2: Site Router Is in Two Areas (One Is Area 0)
This approach makes the Site Router (Router A) an area border router (ABR) under failure. It does work; however, it is not considered an acceptable design because it would make the Site Router part of area 0. This design would require more resources than would be cost-effective in all but the smallest networks (see Figure 6-11).
Design Scenario #3: Site Router Is in One Area
This approach is the most suitable and works even if the backup server (C in Figure 6-12) is located elsewhere. The secret is that C does not summarize for its attached areas; thus, more specific prefixes are originated by C for those sites in failure. The disadvantage is that dedicated backup interfaces are required for each area.
The following are some sample configurations for this design scenario:
Site Router A Configuration
Interface ethernet 0 ip address 18.104.22.168 255.255.255.224 interface serial 0 ip address 22.214.171.124 255.255.255.252 backup interface serial 1 backup delay 0 5 interface serial 1 ip address 126.96.36.199 255.255.255.252 router ospf 1 network 188.8.131.52 0.0.255.255 area 1
Site Router B Configuration
interface fddi 0 ip address 184.108.40.206 255.255.255.248 interface serial 0 ip address 220.127.116.11 255.255.255.252 router ospf 1 network 18.104.22.168 0.0.3.255 area 1 area 1 range 22.214.171.124 255.255.252.0 network 126.96.36.199 0.0.255.255 area 0
Site Router C Configuration
interface fddi 0 ip address 188.8.131.52 255.255.255.248 interface serial 0 ip address 184.108.40.206 255.255.255.252 router ospf 1 network 220.127.116.11 0.0.3.255 area 1 network 18.104.22.168 0.0.255.255 area 0