OSPF: Open Shortest Path First

Learning objective

Explain how OSPF routers form adjacencies, flood link-state advertisements, and compute shortest paths per RFC 2328.

What OSPF is

OSPF (Open Shortest Path First) is the most widely deployed interior gateway protocol, standardized for IPv4 in RFC 2328 (OSPFv2; OSPFv3 for IPv6 is RFC 5340). Unlike RIP's distance-vector design, OSPF is a link-state protocol: every router floods a description of its own directly connected links to every other router in the same area, so all routers end up holding an identical link-state database (LSDB). Each router then independently runs Dijkstra's shortest-path algorithm against that shared database to compute its own best routes. Every OSPF router is identified by a 32-bit Router ID, conventionally written in dotted-decimal form even though it is not necessarily a real IP address.

OSPF cost and the link-state database

OSPF's metric is cost, calculated per interface as reference bandwidth ÷ interface bandwidth, and summed along the whole path. With the common default reference bandwidth of 100 Mbps, a 100 Mbps link costs 1 and a 10 Mbps link costs 10 — faster links are cheaper. Because every router in an area holds the same link-state database, cost comparisons and the resulting shortest-path tree are consistent across the whole area; there is no risk of the reachability disagreements that plague distance-vector protocols.

OSPF packet types

OSPF runs directly over IP (IP protocol 89, no UDP or TCP header) and defines five packet types: Hello (type 1, discovers and maintains neighbors), Database Description / DBD (type 2, summarizes LSA headers during adjacency formation), Link State Request / LSR (type 3, asks a neighbor for a specific LSA in full), Link State Update / LSU (type 4, carries one or more complete LSAs), and Link State Acknowledgment / LSAck (type 5, confirms receipt of an LSU). Every one of these packet types plays a role in the adjacency-formation sequence covered next.

Forming an OSPF adjacency

Two OSPF routers move through a well-defined sequence of neighbor states on their way to a full adjacency: Down → Init (a Hello has been heard, but the local router isn't listed yet) → 2-Way (both routers see each other's Router ID in a Hello) → ExStart (negotiating which router will be master for the exchange) → Exchange (trading DBD summaries of the LSDB) → Loading (requesting and receiving any LSAs not yet held in full) → Full (LSDBs synchronized). On a multi-access network (such as Ethernet) with more than two routers, OSPF also elects a Designated Router (DR) and Backup Designated Router (BDR) so that routers form full adjacencies with the DR/BDR rather than with every other router, cutting down on redundant flooding.

Interactive OSPF adjacency

Step through two routers forming a full adjacency over a point-to-point link, from the first Hello through Database Description negotiation to a synchronized link-state database.

OSPF areas and LSA types

OSPF scales by dividing a network into areas; Area 0, the backbone, must exist, and every other area normally attaches directly to it, which limits how far a single topology change has to propagate. Different Link-State Advertisement (LSA) types carry different information: Type 1 (Router LSA) describes a router's own links within an area; Type 2 (Network LSA) describes a multi-access segment's attached routers, originated by the DR; Type 3 (Summary LSA) lets an Area Border Router (ABR) advertise routes between areas; Type 4 (ASBR Summary LSA) advertises the location of an Autonomous System Boundary Router (ASBR); and Type 5 (AS External LSA) carries routes redistributed from outside OSPF entirely, originated by an ASBR.