OSPF Network Types Explained (Broadcast, Point-to-Point, NBMA)

OSPF network types are an interface-level setting that tells the router how OSPF should behave on that particular medium: whether to elect a designated router (DR) and backup designated router (BDR), whether hellos go out as multicast or unicast, which hello and dead timers apply by default, and whether neighbors are discovered automatically or must be configured by hand. Every OSPF-enabled interface has a network type whether you set one or not - IOS picks a default based on the interface's layer 2 encapsulation. This article is part of our full OSPF guide, and every piece of device output below comes from a live Cisco Modeling Labs topology running IOS XE 17.18.

The setting matters more than it first appears. Two routers with mismatched network types can fail to become neighbors at all (when the types use different default timers), or worse, reach FULL and still fail to install each other's routes (when the timers happen to match but the DR expectations do not). The network type also controls a quirk every CCNA candidate trips over at some point: why loopback interfaces show up in remote routing tables as /32 host routes no matter what mask you configure on them.

The five network types at a glance

• Broadcast - the default on Ethernet. DR/BDR election, multicast hellos to 224.0.0.5, hello/dead timers of 10/40 seconds.
• Point-to-point - the default on serial interfaces running HDLC or PPP. No DR/BDR election, multicast hellos, timers 10/40. Very commonly configured manually on Ethernet links that connect exactly two routers.
• Non-broadcast (NBMA) - the default on Frame Relay physical interfaces. DR/BDR election, unicast hellos, neighbors defined manually with neighbor statements, timers 30/120.
• Point-to-multipoint - never a default; you configure it. No DR/BDR election, multicast hellos, timers 30/120. OSPF treats the segment as a collection of point-to-point links and advertises host routes for the endpoints.
• Point-to-multipoint non-broadcast - the same routing behavior as point-to-multipoint, but with unicast hellos and manual neighbor statements, for clouds that cannot carry multicast at all.

Loopbacks get a sixth, internal type (LOOPBACK) that you never configure directly. It is the reason for the /32 behavior covered near the end of this article.

Broadcast: the Ethernet default

In the lab, R1 and R2 share a LAN in area 0 (10.0.12.0/24), and R2-R3 is a /30 point-to-point wire in area 1. Here is R2's LAN interface with everything left at defaults:

R2# show ip ospf interface Ethernet0/0
Ethernet0/0 is up, line protocol is up
  Internet Address 10.0.12.2/24, Interface ID 2, Area 0
  Attached via Network Statement
  Process ID 1, Router ID 2.2.2.2, Network Type BROADCAST, Cost: 10
  Transmit Delay is 1 sec, State DR, Priority 1
  Designated Router (ID) 2.2.2.2, Interface address 10.0.12.2
  Backup Designated router (ID) 1.1.1.1, Interface address 10.0.12.1
  Timer intervals configured, Hello 10, Dead 40, Wait 40, Retransmit 5
    Hello due in 00:00:01
  Neighbor Count is 1, Adjacent neighbor count is 1
    Adjacent with neighbor 1.1.1.1  (Backup Designated Router)

Four lines carry the story. Network Type BROADCAST is the default IOS chose for Ethernet. State DR means this interface won the DR election (R2 became DR, and the next line shows R1 as the BDR with its interface address). The timer line confirms the broadcast defaults of Hello 10 and Dead 40. On a two-router LAN the DR election buys you nothing, but on a segment with ten routers it collapses what would be 45 full adjacencies into a hub-and-spoke around the DR, and it lets a single type 2 (network) LSA describe the whole segment.

The election itself, priorities, and why the DR does not preempt are covered in our DR/BDR deep dive.

Point-to-point: no DR, faster adjacency

A point-to-point type skips the DR/BDR election entirely, because a link with exactly two routers has nothing to optimize. That removes the wait time for the election and removes the type 2 LSA. To convert an Ethernet link, configure both ends:

interface Ethernet0/1
 ip ospf network point-to-point

In the lab we applied that to both sides of the R2-R3 /30 link. The interface now reports:

R2# show ip ospf interface Ethernet0/1
Ethernet0/1 is up, line protocol is up
  Internet Address 10.0.23.2/30, Interface ID 3, Area 1
  Process ID 1, Router ID 2.2.2.2, Network Type POINT_TO_POINT, Cost: 10
  Transmit Delay is 1 sec, State POINT_TO_POINT
  Timer intervals configured, Hello 10, Dead 40, Wait 40, Retransmit 5
  Neighbor Count is 1, Adjacent neighbor count is 1
    Adjacent with neighbor 3.3.3.3

Notice the state is simply POINT_TO_POINT, not DR or BDR, and there are no Designated Router lines at all. The neighbor table reflects the same thing:

Neighbor ID     Pri   State           Dead Time   Address         Interface
3.3.3.3           0   FULL/  -        00:00:34    10.0.23.1       Ethernet0/1

FULL/ - means fully adjacent with no DR role on the link, and the priority shows 0 because it is meaningless here. If you are used to seeing FULL/DR and FULL/BDR, the dash is not an error - it is the signature of a point-to-point link. (The walk from DOWN through EXSTART to FULL is the same on every type; see OSPF neighbor states for that sequence.)

NBMA and point-to-multipoint

Non-broadcast multi-access (NBMA) exists for media like classic Frame Relay and X.25: many routers share one subnet, but the cloud cannot deliver multicast. OSPF still wants a DR on this type, and since hellos cannot be multicast, you must list the neighbors yourself and force the hub to win the election:

interface Serial0/0
 ip ospf network non-broadcast
 ip ospf priority 100
!
router ospf 1
 neighbor 10.1.1.2
 neighbor 10.1.1.3

The trap with NBMA is DR placement: in a hub-and-spoke cloud, the DR must be the hub (spokes cannot reach each other directly, so a spoke DR breaks flooding). Setting ip ospf priority 0 on the spokes is the standard defense.

Point-to-multipoint sidesteps the whole problem. It tells OSPF to treat the cloud as a bundle of point-to-point links: no DR, no election games, host routes advertised for every endpoint so spoke-to-spoke traffic relays through the hub correctly:

interface Serial0/0
 ip ospf network point-to-multipoint

You will rarely touch either type on modern hardware, but they survive in two places: legacy WAN migrations, and DMVPN or other overlay hub-and-spoke designs where point-to-multipoint is still a legitimate choice. There is no lab capture for these two here because the CML topology is all Ethernet - which is exactly the point: on Ethernet you choose between broadcast and point-to-point, and that decision covers 95 percent of real deployments.

Changing the network type (and how mismatches bite)

The command is always the same, applied per interface:

interface Ethernet0/1
 ip ospf network point-to-point

The rule that goes with it: change both ends of the link. Network types with different default timers (broadcast at 10/40 versus point-to-multipoint at 30/120, for example) will refuse to become neighbors, because the hello packet carries the hello and dead intervals and the receiver rejects any hello that disagrees. When we deliberately created a hello disagreement on the R2-R3 link in the lab, debug ip ospf hello on R2 printed the classic line:

*Jul  4 23:31:44.787: OSPF-1 HELLO Et0/1: Mismatched hello parameters from 10.0.23.1

That is the exact symptom you get from a broadcast versus point-to-multipoint type mismatch. The nastier case is broadcast on one end and point-to-point on the other: both use 10/40, so the hellos pass inspection and the neighbors reach FULL, but one router describes the link with a network LSA and DR while the other describes a point-to-point connection. The LSDB entries do not line up, and routes across that link fail the SPF check even though the neighbor table looks perfect. If show ip ospf neighbor says FULL but the routes are missing, comparing show ip ospf interface on both ends should be an early step. (More neighbor-formation failure patterns are in troubleshooting OSPF neighbors not forming.)

Which type should you use?

For links between exactly two routers, point-to-point is the right answer even on Ethernet: no election delay, no type 2 LSA, one less moving part. For true multi-access LANs with three or more routers, leave broadcast alone. And for loopbacks, point-to-point fixes the mask problem, which the lab shows nicely.

R3 carries four loopbacks numbered 172.16.0.1/24 through 172.16.3.1/24 in area 1. With default settings, here is what R1 learns:

R1# show ip route ospf
      2.0.0.0/32 is subnetted, 1 subnets
O        2.2.2.2 [110/11] via 10.0.12.2, 00:00:43, Ethernet0/0
      3.0.0.0/32 is subnetted, 1 subnets
O IA     3.3.3.3 [110/21] via 10.0.12.2, 00:00:36, Ethernet0/0
      10.0.0.0/8 is variably subnetted, 3 subnets, 3 masks
O IA     10.0.23.0/30 [110/20] via 10.0.12.2, 00:00:43, Ethernet0/0
      172.16.0.0/32 is subnetted, 4 subnets
O IA     172.16.0.1 [110/21] via 10.0.12.2, 00:00:36, Ethernet0/0
O IA     172.16.1.1 [110/21] via 10.0.12.2, 00:00:36, Ethernet0/0
O IA     172.16.2.1 [110/21] via 10.0.12.2, 00:00:36, Ethernet0/0
O IA     172.16.3.1 [110/21] via 10.0.12.2, 00:00:36, Ethernet0/0

Every loopback arrives as a /32 host route, mask ignored, because the LOOPBACK network type advertises a stub host route regardless of the configured mask. Apply ip ospf network point-to-point under each loopback on R3, and R1 now sees the real masks:

R1# show ip route ospf | begin 172.16
      172.16.0.0/24 is subnetted, 4 subnets
O IA     172.16.0.0 [110/21] via 10.0.12.2, 00:00:16, Ethernet0/0
O IA     172.16.1.0 [110/21] via 10.0.12.2, 00:00:16, Ethernet0/0
O IA     172.16.2.0 [110/21] via 10.0.12.2, 00:00:15, Ethernet0/0
O IA     172.16.3.0 [110/21] via 10.0.12.2, 00:00:15, Ethernet0/0

Same loopbacks, now advertised as the /24 networks that were actually configured. If you use loopbacks to simulate LANs in a lab, or to anchor services on real networks, this one-liner is the difference between advertising the subnet and advertising a single host. (The [110/21] metric in both outputs is plain OSPF cost arithmetic; see how OSPF cost works if those numbers are unfamiliar.)

Key Takeaways

• The network type is per interface and controls DR/BDR election, hello delivery (multicast or unicast), default timers, and neighbor discovery.
• Ethernet defaults to broadcast (DR/BDR, 10/40); serial HDLC/PPP defaults to point-to-point (no DR, 10/40); NBMA and point-to-multipoint default to 30/120.
• FULL/ - in the neighbor table is normal on point-to-point links - it means no DR exists, not that something failed.
• Match the type on both ends. Different default timers block the adjacency outright; broadcast against point-to-point forms FULL but breaks routing through mismatched LSAs.
• Use point-to-point on any two-router Ethernet link, and on loopbacks whenever you want the real mask advertised instead of a /32.