What Is OSPF?

An Interior Protocol That Maps the Network

OSPF — Open Shortest Path First — is an interior gateway protocol, meaning it routes traffic within a single network or autonomous system, not between separate organizations. That outer job belongs to BGP; OSPF handles the internal paths, a division covered in IGP vs EGP.

What makes OSPF distinctive is that it is a link-state protocol. Rather than each router knowing only "to reach network X, hand the packet to neighbor Y," every OSPF router learns the entire topology — every router, every link, and the state of each — and stores it in a link-state database. With a full map in hand, a router can compute the genuinely shortest path to anywhere, instead of trusting secondhand directions from neighbors. This whole-map approach is the source of OSPF's speed and intelligence.

How Routers Build a Shared Map

OSPF assembles that map through a sequence of steps that every router participates in:

1. Discover neighbors — routers send Hello messages to find other OSPF routers on their links and establish adjacencies.
2. Advertise links — each router describes its own connections in Link-State Advertisements (LSAs) — essentially "here are the links I have and their costs."
3. Flood the LSAs — these advertisements are flooded throughout the area so every router receives every other router's LSAs.
4. Build the database — each router assembles all the LSAs into an identical link-state database, a complete picture of the topology.
5. Run SPF — each router runs the shortest-path-first calculation on that database to find the best route to every destination, and installs the results in its routing table.

Because every router works from the same database, they all reach a consistent set of best paths — there is no disagreement about the shape of the network. When a link fails or appears, the affected router floods an updated LSA, every router updates its database, and all of them recompute. This is why OSPF converges — settles on new correct routes — quickly after a change.

Cost: How OSPF Picks the Shortest Path

The "shortest" path in OSPF is not the fewest hops — it is the lowest total cost. Each link is assigned a cost, conventionally derived from its bandwidth, so faster links have lower cost and are preferred. A router sums the costs along each possible path and chooses the cheapest.

This matters enormously compared to counting hops. Imagine two paths to a destination: one crosses a single slow link, the other crosses two fast links. A hop-counting protocol would pick the single slow hop; OSPF, weighing bandwidth, correctly prefers the two fast links because their combined cost is lower. The algorithm doing this calculation is Dijkstra's Shortest Path First — a classic graph algorithm that, given the link-state map and the costs, finds the minimum-cost route to every node. The map plus Dijkstra is the entire intelligence of OSPF.

Areas: Keeping It Scalable

Flooding every LSA to every router and recomputing the whole map works beautifully on a modest network — but on a very large one it would mean enormous databases and constant recalculation. OSPF solves this with areas, which divide the network into regions.

Within an area, routers keep full topology detail. Between areas, only summarized information is exchanged, and all areas connect through a central backbone called area 0. The effect is containment: a topology change in one area is recalculated only within that area, and other areas see just a summary update rather than the full churn. This keeps each router's database and computation bounded no matter how large the overall network grows, and it is the main reason OSPF scales to big enterprise and provider networks.

OSPF vs RIP, and Where It Fits

OSPF's design is best appreciated against the protocol it largely replaced. RIP is a distance-vector protocol: a router knows only how many hops away a destination is and shares its routing table with neighbors, who pass it along like a game of telephone. RIP converges slowly, can loop during changes, and caps at 15 hops. OSPF, by contrast, gives every router the full map and a cost-based calculation:

In practice OSPF runs inside an autonomous system while BGP connects that system to the rest of the internet — the two complement each other rather than compete. OSPF finds the best internal path; BGP decides which neighboring network to hand external traffic to. Together they cover the full journey of a packet from a local device out to a distant server.

Frequently Asked Questions

What is OSPF?

Open Shortest Path First — a link-state interior gateway protocol that routes within a network or autonomous system. Each router learns the full topology via link-state advertisements, then computes the shortest path to every destination with the SPF algorithm. It is the most widely deployed interior routing protocol.

How does OSPF choose the best path?

It assigns each link a cost, usually based on bandwidth, and picks the lowest-total-cost path. Every router builds an identical map and runs Dijkstra's shortest path first algorithm, so all reach a consistent view. Higher-bandwidth links have lower cost and are preferred.

What is the difference between OSPF and RIP?

RIP is distance-vector, knowing only hop counts and sharing its table with neighbors, converging slowly and capped at 15 hops. OSPF is link-state — every router learns the whole topology and computes paths by cost. OSPF converges faster, scales larger, and chooses smarter paths, so it largely replaced RIP.

What are OSPF areas?

Regions that divide a large OSPF network for scalability. Routers keep full topology detail only within their area and exchange summaries between areas through a backbone area 0. This bounds each router's calculation and limits how far changes ripple, letting OSPF scale.

Is OSPF still used today?

Yes — it is one of the most common interior routing protocols, standard in enterprises, data centers, and many providers for routing within an autonomous system. It works alongside BGP, which handles routing between autonomous systems.