Continuing on with our detailed discussion of OSPF, let’s look at how OSPF routers handle prefixes from external routing domains. Refer to Figure 1:
Let’s say that there are 100 subnets in the RIP cloud. If we configure R3 to redistribute the RIP routes into OSPF (using the “redistribute” command under OSPF), R3 will be an ASBR (Autonomous System Boundary Router). Likewise, let’s assume that there are also 100 prefixes within the EIGRP cloud, and that we have configured R4 as an ASBR, redistributing EIGRP into OSPF. As part of their ASBR function, R3 and R4 will advertise each individual prefix within their respective external routing domain into Area 0, using a Type 5 (external) LSA. Since by default Type 5 LSAs have autonomous system flooding scope, they will be passed into Area 1 by R3 and R4. Thus, R5 will have two possible paths (via R3 or R4) to all 200 external prefixes.
When configuring an ASBR, we have the option of including or ignoring internal costs (those of the links within the OSPF cloud) when calculating the metrics to external destinations. Let’s assume that “O E1” (“OSPF External Type 1”) routes are used, for which the internal metrics are considered. If we look at things from R5’s perspective, each of the destinations listed would be reached using the lowest-cost path, via the specified next hop:
- Subnet A via R3 (cost = 3)
- Subnet B via R3 (cost = 3)
- Subnet C via R4 (cost = 3)
- Subnet D via R4 (cost = 3)
- RIP cloud via R3 (100 prefixes, each at cost = 21)
- EIGRP cloud via R4 (100 prefixes, each at cost = 21)
At this point, each router within the OSPF routing domain will know the best path to each individual prefix within the OSPF cloud, as well as to each individual prefix within the RIP and EIGRP clouds. The best path to each external prefix will appear in R5’s IP routing table as “O E1” route.
Next time, we’ll take a detailed look at OSPF’s route summarization features, including possible unintended consequences of their use.
Author: Al Friebe