I'm working on optimizing an OSPF network, and I'm looking at different network types. Here is what I understand:

  • Broadcast: Dynamic neighbors; uses multicast; elects DR
  • NBMA: Static neighbors; uses unicast; elects DR
  • PTP: Single neighbor; uses multicast; does not elect DR
  • PTMP: Dynamic neighbors; uses multicast; does not elect DR

How does PTMP maintain efficiency without electing a DR? I assume it uses the "hub", but how does it know what the hub is if everything is on the same subnet?


Routers form FULL adjacency with each other. This mode is useful for hub and spoke design. One of the examples - Frame Relay network.

  • Please clarify what this means. Apr 25 '17 at 13:09
  • When network is broadcast your routers at the shared segment will form FULL adjacencies with DR and BDR only. So DR is responsible for event propogation. When network is p2mp all the routers form full adjacencies and send updates directly to their neighbors.
    – ar_
    Apr 25 '17 at 13:17
  • Ah, I see. One last thing, could you please explain why this is advantageous over having a DR? Because it seems that this just makes it less efficient on terms of route flooding. Apr 25 '17 at 13:24
  • Well, I'm not sure p2mp is advantage. Many things depend on topology. For some p2mp is good. As for me, I'm always trying to create p2p only topology if possilbe for the sake of scalability, convergance and simplicity. DR is actually some kind of hack, that makes mesh topology look like star with DR in the center. So p2p seem more clear and elegant, when it is applicable.
    – ar_
    Apr 25 '17 at 13:30
  • Interesting. According to Mikrotik, PTMP is the most robust network type. Apr 25 '17 at 13:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.