As you must be knowing that multiple routers can exist in same network segment (broadcast network) so in order to minimize the amount of information exchange on a particular segment ospf elects DR and BDR. The idea behind this is that routers have a central point of contact for information exchange. Instead of each router exchanging updates with every other router on the segment, every router exchanges information with the DR and BDR. The DR and BDR relay the information to everybody else.
Now point to point networks as name implies does not have multiple connections. OSPF will always form an adjacency with the neighbor on the other side of a point-to-point interface so there is no concept and need of DR or BDR.
To know how cost is calculated please read this link
*Pls rate all helpfull post
point-to-point networks has just 2 neighbors, and thus no need for DR/BDR (they are required on multiaccess networks (broadcast and non-broadcast) which will have full mesh neighborships if not using DR/BDR which means alot of routers resources + huge LSA floods).
For OSPF cost = 10^8/interfaceBW (the generalized form is ReferenceBW/InterfaceBW) - Cost = The outgoing cost for packets transmitted from this interface (used when executing the SPF tree by adding it to the cost of relevant path to the destination - When talking about this issue in the essence of the received routing updates, then the cost of the interface that the route was received from is added to the cost declared in the LSA in order to calculate the total cost.
Thanks for the information provided.Can you please explain me what is the difference in
Broadcast and Non Broad Cast network in ospf. Does DR/BDR required in point- multi point.
Read this link and it will clear your doubts. Incase if not please come back again with your doubt.
*Pls rate all helpfull post
Point-to-multipoint is treated as a collection of point-to-point links and thus no DR/BDR is required.
Broadcast: OSPF routers on broadcast networks will elect a DR and a BDR (since it is multiaccess) - OSPF packets are multicast.
NBMA: Routers will elect DR and BDR (since it is multiaccess), but since it is a non-broadcast, routers will have to communicate via unicast rather than multicast.
I believe that it is helpful to look at each of the terms and understand what they mean. We generally describe an interface as point to point or as multiaccess. The distinction is pretty simple: on a point to point connection we have an interface and through that interface we are able to communicate with one and only one device. A multiaccess interface is an interface through which we may communicate with more than one device.
A multiaccess interface may be described as broadcast or as non-broadcast. In a broadcast interface we are able to send a single frame and that frame is seen by multiple devices. In a non-broadcast interface we are not able to send a single frame that will be seen by multiple devices.
Having looked at the meaning of the terms lets look at some examples and how OSPF works in them. A point to point interface might be a serial interface with HDLC or PPP encapsulation or it might be a point to point subinterface Frame Relay. In point to point interfaces OSPF does not elect Dr or BDR because there is no benefit to them in this environment. A multiaccess broadcast interface might be an Ethernet interface. In multiaccess broadcast interfaces OSPF will elect a DR and a BDR because they potentially simplify the process of forming adjacencies. A non-broadcast multiaccess interface might be the physical serial interface with Frame Relay (possibly with multiple PVCs) or a multipoint subinterface with Frame Relay. In this situation if the router sent a single frame it could send the single frame on only a single PVC and thus is not capable of true broadcast capability. What the router can do is to make copies of the frame and send a copy on each PVC. But this is not quite the same thing as having true broadcast capability. In non-broadcast multiaccess OSPF will elect a DR and a BDR because they potentially simplify the process of forming adjacencies.
As Mohammed points out point to multipoint is a special case of NBMA. Physically and logically it is NBMA but OSPF is defined to have different behaviors. There is no election of DR/BDR and every router forms a full adjacency with every neighbor that it finds.
Thanks Everybody for helping me in understanding these concepts.
Can any one help me in giving few OSPF trouble shooting scenarios from any good link.
Any help will be highly appreciated.
I am glade that we have helped you out, here you are a nice document for Troubleshooting OSPF:
Hi Rick and all,
I hope this thread finds you well. In your last few paragraph
What the router can do is to make copies of the frame and send a copy on each PVC. But this is not quite the same thing as having true broadcast capability. In non-broadcast multiaccess OSPF will elect a DR and a BDR because they potentially simplify the process of forming adjacencies. As Mohammed points out point to multipoint is a special case of NBMA. Physically and logically it is NBMA but OSPF is defined to have different behaviors. There is no election of DR/BDR and every router forms a full adjacency with every neighbor that it finds.
Does that meant ->
In NBMA network with 4 routers with OSPF configured -> I will have 3 neighbours and 2 adjacency on a DROTHER router.
In a point to multipoint setup with the same amount of routers -> I will have 3 neighbours and 3 adjacency on a DROTHER router. ?
With DR / BDR in a nbma network with 4 routers we will see 3 neighbors on one router
on DR we will see x1 BDR and x2 DROTHER
on BDR we will see x1 DR and x2 DROTHER
on DROTHER we will see x1 DR x1 BDR x1 DROTHER
The states to the DR and BDR and from the DR and BDR will be FULL. The Drothers will be 2WAY with each other. Now as you may know, the OSPF update behavior is via DR and BDR, and then the DR updates the rest of the DROTHERS.
With Point to multipoint networks, are a special configuration of NBMA networks in which the networks are treated as a collection of point to point links. Routers on these networks do not elect a DR and BDR and because the networks are seen as point to point links, the OSPF packets are multicast. This is assuming the transit links support broadcast and multicast. If not then there is a option to choose/configure point to multipoint non broadcast where we would have to statically define our neighbor.
As a result its like having point to point neighborships with FULL state and updates go out to everyone.
Thank you for your attention on this thread.
Nice pointing out on multipoint, i forget they are suppose to treat each connection as point to point.
Hence assuming if router A,B,C,D can be physically connected in full mesh, this means each router would have have 3 neighbors and 3 full adjacency - am i right ?
Yes - absolutely correct, seems though you are getting a good grip ;)
Some additional notes on OSPF costing.
If I remember correctly, OSPF, strictly speaking, doesn't require bandwidth to map into the cost metric. However, as Mohammed notes, Cisco (starting with IOS 10.3) assigns a cost of 1 to links with 100 Mbps. Not all vendors do this, or use the same base bandwidth. If you mix vendor equipment, you might want to insure all costing is playing by the same rules.
Second, if you use 100 Mbps Etherchannel, gig or 10 gig links, by default, they too get a cost of 1. This means OSPF can not see the difference between links 100 Mbps or better. This can be dealt with by moving the base bandwidth metric to represent the highest speed, using the command "auto-cost reference-bandwidth mbps", or by setting the cost on every interface manually.
Lastly, if you have network that's deep enough with low bandwidth links while running on a higher base reference bandwidth, the cost counter fills. Once again, OSPF will be unable to determine which path is better.