Questions: OSPF: Open Shortest Path First

Question 1

A link fails between two routers in OSPF Area 3. The network has 5 areas total (Area 0 through Area 4). Which routers must re-run Dijkstra's algorithm?

• A. All routers in the network, because OSPF floods LSAs to every router regardless of area
• B. Only the two routers directly adjacent to the failed link
• C. All routers within Area 3, plus Area Border Routers that summarize Area 3 routes to other areas
• D. No routers — OSPF pre-computes backup paths and switches automatically without recalculation

Answer: (C)

OSPF areas limit the propagation of detailed topology changes. A link failure in Area 3 triggers LSAs that flood only within Area 3, so only routers inside Area 3 re-run Dijkstra on the updated LSDB. Area Border Routers (ABRs) may generate summarized LSAs to inform other areas of changed summary routes, but routers in Areas 1, 2, and 4 never see the internal topology detail. This is the core scalability benefit of areas: failures are localized, minimizing unnecessary computation across the network.

Question 2

OSPF discovers two routes to the same destination with total costs of 20 and 20 (identical). What does OSPF do?

• A. Drops one route arbitrarily and installs only the other to avoid routing loops
• B. Installs both routes and load-balances traffic across them (ECMP)
• C. Waits for the costs to differ before making a forwarding decision
• D. Selects the route through the router with the lower Router ID as a tiebreaker

Answer: (B)

OSPF supports Equal-Cost Multipath (ECMP) routing: when Dijkstra finds multiple paths with the same total cost, all are installed in the routing table and traffic is load-balanced across them. This is a significant advantage over protocols that must pick one best path — ECMP improves throughput and resilience. Option D describes a tiebreaker used in some other contexts (like Designated Router election), but it does not apply to ECMP path selection.

Question 3

In OSPF, a central route-computation server calculates shortest paths and distributes them to most routers.

• T. True
• F. False

Answer: False

OSPF is fully distributed. Every router independently builds the same complete Link-State Database (LSDB) through the flooding of LSAs, and every router independently runs Dijkstra's algorithm on that database to compute its own routing table. There is no central authority. The key insight is that because all routers share the same LSDB, they all compute consistent, loop-free results without coordination — each router reaches the same conclusion autonomously.

Question 4

Area 0 (the backbone area) must connect all other OSPF areas, either directly or through virtual links.

• T. True
• F. False

Answer: True

OSPF's hierarchical area design requires all non-backbone areas to connect to Area 0. Inter-area routing flows through the backbone: when traffic moves from Area 2 to Area 4, it goes through Area 0. Area Border Routers (ABRs) connect their area to Area 0 and summarize topology between them. If a non-backbone area cannot connect directly to Area 0, OSPF provides virtual links as a workaround — but the logical requirement that all routing pass through Area 0 remains absolute.

Question 5

Why does OSPF organize routers into areas rather than flooding all topology information to every router in the network?

Model answer: Areas limit the scope of topology flooding and path computation. In a flat OSPF network, every link change floods LSAs to every router, and every router must re-run Dijkstra on the entire network graph. As the network grows, this becomes prohibitive: LSA traffic consumes bandwidth, and Dijkstra re-runs consume CPU proportional to the number of edges. Areas contain topology detail within logical boundaries — a failure in one area triggers flooding and re-computation only within that area. Area Border Routers summarize inter-area reachability, so routers in other areas see only abstract route changes, not internal topology. The result is smaller LSDBs, faster convergence, and dramatically reduced overhead in large networks.

The tradeoff is that summarization loses some information — ABRs cannot propagate every metric detail across area boundaries, and OSPF cannot achieve globally optimal paths in all cases with summarization. But the scalability gains in large networks far outweigh this cost.