In the topology shown, the 10.10.0.0/16 network is redistributed into two different OSPF processes—OSPF 1 and OSPF 2—from router R3. The redistribution uses the external route types:
Into OSPF 1 as an E1 route with a metric of 100
Into OSPF 2 as an E2 route with a metric of 100
When redistribution occurs in OSPF, two types of external routes can be injected:
Type E1: Adds the internal OSPF cost to the external cost (total path cost = external metric + internal OSPF path)
Type E2: Considers only the external cost and ignores internal path cost (default behavior)
In this case:
The path via R2–R3 (E1) includes both the external cost and the internal cost from R1 to R2 to R3. That is: 10 (R1–R2) + 10 (R2–R3) + 100 (external) = 120.
The path via R4–R5–R3 (E2) will show a constant metric of 100 regardless of the internal OSPF path (10 + 10 = ignored).
So when the link between R2 and R3 fails, the path R1–R4–R5–R3 (E2 route with constant metric 100) becomes the preferred route.
However, when the link between R2 and R3 is restored, the E1 route will be recalculated as:
OSPF always prefers E1 routes over E2 routes when both are available, even if the E2 route appears to have a better metric. This is because E1 routes provide more accurate end-to-end cost calculation.
Therefore, upon restoration, the routing table reverts to the E1 route via R1–R2–R3.
This behavior is aligned with OSPF protocol standards and is emphasized in CCDE v3.1 under the “Protocol Design Implications” domain, which focuses on redistribution behaviors, route-type preferences, and convergence consistency in multi-domain IGP designs.
