3 OSPF Fundamentals and Configuration

How OSPF calculates routes

Open Shortest Path First (OSPF) is a link-state interior gateway protocol. Routers share information about links and networks, build a link-state database (LSDB), and independently calculate shortest paths with the Shortest Path First (SPF) algorithm. The resulting best routes are installed in the routing table when selected and usable.

OSPF supports equal-cost paths, so a router can use multiple routes when their costs are equal. For a path, the total cost is the sum of its outgoing interface costs:

Here, $c_i$ represents the cost of an outgoing interface on the path. OSPF prefers the path with the lowest total cost. The exact default cost calculation can depend on the vendor and configuration.

Areas and router identity

An area is a set of OSPF-connected links and routers that share an area-specific LSDB. In a single-area network, all participating interfaces belong to the same area, commonly Area 0.

The OSPF process ID identifies a local process on a router and does not need to match the process ID on a neighboring router. The router ID is a 32-bit identifier. Explicitly configuring a unique, stable router ID makes identification and troubleshooting easier.

Neighbors and adjacencies

Routers send OSPF Hello packets to discover and maintain neighbors. Once bidirectional communication is confirmed, routers form adjacencies where required and synchronize their LSDBs. They then flood link-state advertisements (LSAs) within the area. Each router runs SPF against its own LSDB to derive routes; OSPF does not simply pass a neighbor’s finished routing table along.

The common neighbor states, in order, are:

• Down — no recent Hello has been received.

• Init — a Hello has been received, but it does not list this router as a neighbor.

• 2-Way — Hellos confirm communication in both directions. On a shared broadcast network, two routers that are neither the DR nor BDR may remain in this state; that can be normal.

• ExStart and Exchange — routers establish database-exchange roles and exchange summaries of their LSDBs.

• Loading — routers request and receive needed LSAs.

• Full — the adjacency databases are synchronized.

On broadcast networks, OSPF elects a Designated Router (DR) and Backup Designated Router (BDR) to reduce the number of adjacencies needed. On point-to-point links, the routers form an adjacency directly.

Basic single-area configuration

The example connects two LANs through a point-to-point transit network. It uses Cisco IOS/IOS XE syntax; interface names may differ by device. Configure interface addresses first, then enable OSPF on interfaces whose addresses match the network statements. A wildcard mask is the inverse of a subnet mask; for example, 0.0.0.255 matches any address in a /24. The OSPF network command both selects interfaces and assigns them to an area.

R1:

R2:

The example uses process 1 on both routers, but process IDs are local and need not match. The two sides of the transit link must use compatible OSPF settings and the same area. Cisco also supports enabling OSPF directly under an interface with ip ospf <process-id> area <area-id>.

Verification and troubleshooting

On Cisco IOS/IOS XE, learned OSPF routes are commonly marked O in show ip route. For example, O 192.168.20.0/24 indicates an OSPF-learned intra-area route. A route appears in the routing table only if it is selected as a best route and is usable.

Use these commands to check the neighbor relationship, OSPF-enabled interfaces, LSDB, and OSPF routes:

A healthy point-to-point adjacency should reach Full, and each router should learn the other router’s LAN prefix. If a neighbor is absent or stuck before Full, troubleshoot in this order:

1. Confirm both interfaces are up and have addresses in the expected connected subnet; test basic IP reachability.

2. Check that OSPF is enabled on the intended interfaces and that both interfaces use the same area and compatible area type.

3. Compare Hello and dead intervals, authentication settings, and network type; check that router IDs are unique.

4. If the neighbor is stuck in ExStart or Exchange, check for an interface MTU mismatch and other database-description exchange issues.

5. Confirm that an interface is not configured as passive if it must form a neighbor relationship. Depending on the configuration, a passive interface can advertise its connected network while suppressing neighbor-forming Hellos.

6. If the neighbor reaches Full but a route is missing, check whether the destination network is included in OSPF, whether its LSA is present in the LSDB, and whether another route is preferred.

Area, subnet, timer, and router-ID mismatches can prevent neighbor formation. An MTU mismatch is a common cause of neighbors remaining in ExStart or Exchange.