Virtual Link Behind a Stub Area
R-A is a backbone router in area 0. R-B is an ABR between area 0 and area 5. R-C is an internal router in area 5 and holds the only link into a newly created area 6, where a recently acquired subsidiary runs 12 OSPF routers. Area 5 was configured as a stub two years ago to keep memory consumption down on two elderly access routers, and nobody has revisited that decision. Traffic from area 6 currently reaches area 0 by going R-C, area 5, R-B, which works, but area 6 is now a third-level area: it has no direct or virtual attachment to the backbone, and R-C's router LSA shows no backbone attachment. Design intent is for R-C to become a genuine ABR for area 6. The two old access routers in area 5 remain in service, and the change window is 30 minutes.
1. Can a virtual link solve this, and what blocks it today?
Read the worked answer
A virtual link through area 5 is the textbook fix and it cannot be built while area 5 is a stub. The transit area must be a normal area carrying a complete LSDB, because the virtual link is modelled as an unnumbered point-to-point link in area 0 whose path is resolved through the transit area's SPF. A stub area's LSDB is deliberately incomplete and cannot carry the transit topology required, and RFC 2328 excludes stub areas as virtual-link transit. So the first step is removing the stub configuration from area 5 on every router in that area inside the same window - a router that keeps stub while a neighbor removes it will have its hellos rejected, which is a hard outage rather than a degraded state. The second step is checking the memory consequence on the two old access routers: removing stub floods type 5 from area 0 and wherever else, which is precisely why the stub existed. Estimate the new LSDB against their free memory before the window, and if headroom is genuinely insufficient, the virtual link is the wrong design and R-C should be attached to area 0 directly.
2. Assuming area 5 is converted, describe the virtual link configuration and the prerequisites for it to come up.
Read the worked answer
Configure the virtual link in the transit area's context on both endpoints: on R-C, in area 0.0.0.5, a virtual link to R-A's router ID; on R-A, in area 0.0.0.5, a virtual link to R-C's router ID. It is symmetric, so if only one end is configured the link stays down. Hello and dead intervals must match between the two virtual-link configurations and are independent of the physical interfaces' values; authentication, if used, is configured on the virtual link itself rather than inherited from the transit interfaces. Prerequisites: both routers must have an interface in area 5, there must be a full-adjacency path between them across area 5, and R-C becomes an ABR only once the virtual link is up. Verify with show router ospf neighbor, where the virtual neighbor appears in area 0, and show router ospf interface for the virtual link state. Note that the virtual link's cost equals the intra-area cost across area 5, so it is never a cheap path and should not be used to carry production transit traffic.
3. The customer wants the least risky outcome. Compare the virtual link against the alternatives and make a recommendation.
Read the worked answer
Recommend avoiding the virtual link if the design allows it. Option A is to merge area 6 into area 5: the subsidiary has 12 routers, which fits comfortably within area 5's SPF and LSDB budget, the third-level area disappears, and no transit configuration is needed. Keeping area 5 as a stub preserves the memory savings on the old access routers, provided the subsidiary has no external redistribution requirement - if it does, a stub cannot serve it and the area type must change anyway. Option B is to give R-C a direct physical link to R-A and place that link in area 0, making R-C a genuine ABR with no virtual link; this is the cleanest outcome if fibre is available. Option C, the virtual link, is correct but operationally fragile: it depends on transit-area reachability, needs matching configuration on two routers, and fails opaquely if area 5 partitions or flaps. Choose A unless the subsidiary needs its own area for policy or scale, choose B if a physical path exists, and choose C only when the area boundary must be preserved and area 5 can safely become a normal area.