Put five routers on one Ethernet segment and naive OSPF would build ten full adjacencies, each pair flooding LSAs to each other in duplicate. OSPF's answer is representative democracy: elect one router — the Designated Router — to speak for the segment, with a Backup DR synchronised and ready. Every other router (DROther) forms full adjacencies only with the DR and BDR, staying politely at 2-Way with its fellow DROthers. Flooding becomes hub-and-spoke: updates go to the all-DR multicast 224.0.0.6, and the DR re-floods to all OSPF routers on 224.0.0.5. The DR also originates the Type 2 network LSA representing the segment itself.
How the election actually behaves
- Highest interface priority wins (default 1); priority 0 means ineligible — the router will never be DR or BDR on that segment.
- Ties break on highest router ID.
- The election is non-preemptive: a new router with better credentials does not displace a sitting DR. On segment bring-up, the BDR is actually elected first and promoted; in a running network, the BDR is promoted when the DR dies and a fresh BDR is elected.
- Consequence: boot order, not configuration, often decides the DR — so a deterministic design sets priorities explicitly (e.g. 100 on the intended DR, 50 on the BDR, 0 on access routers) and, when it matters, restarts OSPF processes in order.
R3(config)# interface gi0/1R3(config-if)# ip ospf priority 100R1# show ip ospf interface gi0/1 | include State|Priority State DROTHER, Priority 1 Designated Router (ID) 3.3.3.3, Interface address 10.1.0.3
The five network types
- Broadcast — the Ethernet default: DR/BDR election, hellos every 10 s, neighbours auto-discovered by multicast.
- Point-to-point — the default on serial and the right choice for routed Ethernet links between two routers: no DR, no Type 2 LSA, fastest possible adjacency.
- Non-broadcast (NBMA) — classic Frame Relay hub-and-spoke: no multicast, so neighbours are configured manually and hellos are unicast; DR election happens and the hub must win it, since only the hub can reach every spoke.
- Point-to-multipoint — treats an NBMA cloud as a bundle of point-to-point links: no DR, endpoints advertised as /32 host routes; the robust choice for partial-mesh WANs.
- Point-to-multipoint non-broadcast — the same, with unicast hellos for clouds without multicast support.
The one command that removes a whole class of problems
Two routers joined by a routed Ethernet link still default to broadcast type and burn a wasteful election, a Type 2 LSA, and a 40-second wait behind the scenes. Declaring the link point-to-point removes all of it:
R1(config)# interface gi0/0R1(config-if)# ip ospf network point-to-point
Network type must match on both ends — mismatches (say, broadcast against point-to-point) can form adjacencies that then fail in subtle ways, because the two sides disagree about whether a DR and its Type 2 LSA should exist. It is one of the classic ENARSI troubleshooting scenarios: the neighbour table looks fine, the database is subtly wrong, and the routes never appear. When a multi-access segment misbehaves, start with show ip ospf interface: it names the network type, the priority, the DR and BDR — the entire cast of this chapter — in ten lines.