Role of DR and BDR in OSPF network (pic).

rjbarlow

Hi all, I have some trouble to contextualize the ospf role of DR and BDR in a Multi-Access Network.



First, the concept of Multi access network.

From what I read, understood that a Multi-Access network is:
1) A network who has multiple devices in it and with devices all in the same subnet.
a) If these devices have the capability of braodcast, the multi accesss network is named Broadcast Multi Access Network (BMA).
b) If the devices in it don't have this capablity, the net is named Non Broadcast Multi Access (NBMA).

Moreover for Multi Access Network is intended a network multi-accessed from inside and not necessarily from outside (this is very important).

Come to the pic.

In this case is represented a BMA, but I think is useless the election of a DR and BDR because it is one only router (router 5), that is phisically attached to the rest of the net and it is the unique that is capable to advertise the entire network routing information for the topology database to router 1, 2, 3 and 4.
In fact if router 5 fails, all other routers of the subnet 137.3.4.0 are cut from the rest of the net, then also if router 4 for example was the BDR, he surely can't do nothing.

Then my conclusion is that always at least 2 routers must to be phisically connected to the rest of the net in order to DR and BDR have sense.

Are correct these considerations?

Thank You and sorry for the length.

dtlokee

I think you're missing the boat here. The concept of a DR and BDR is not related to a "primary path" and a "backup path" as it seems you have described. The DR and BDR are used on multiaccess media (like Ethernet) to reduce the number of Full OSPF adjacencies that are required. In a multiaccess network there will be 2 routers, one designated as the DR and one desiginated as the BDR and any other routers will become DROTHER. All the routers on the multiaccess segment will establish a full adjacency with both the DR and BDR. All DROTHER routers will only establish a 2 way relationship with other DROTHER routers. So if you had an Ethernet segment with 10 routers, there will only be 17 full adjacencies with the use of a DR/BDR, vs a total of 45 if they were fully meshed(without the DR/BDR).

When you say Broadcast multi-access, or non-broadcast multi-access you are referring to the L2 topology, not the layer3 topology. Ethernet is an example of broadcast multiaccess, and Multi-point frame-relay would be an example of non-broadcast multi-access.

mikearama

I think you're correct, rj... that's not the right kind of diagram to properly show DR's and BDR's, if only because router5 is a single point of failure.

By the way, were you told that router5 was the DR? Otherwise, based on priority or highest IP, it might not be.

If it is, then it doesn't matter which router is the BDR... router5 goes down, and the two 'networks' have lost connectivity. The BDR will still become the DR and the receive the LSA's for it's portion of the network, but...

if Router5 was the DR, and let's pretend that router4 was the BDR... as soon as router5 goes down, Router4 will promote to DR. At the same time, routers 6 - 8 will not have any designated routers, and will elect for themselves a DR and a BDR.

Tricky one.

Netstudent

I don't even see one OSPF broadcast network in that diagram. Those are all point to points right? IF you want to see the election, you need to put at least 2 routers on the same switch. Or as others stated before me, they ned to be on the same "segment" and the same subnet. those ethernet links in the diagram would still need their own subnet to route the packets. IF all those routers were in the same subnet, the routers wouldn't know where to route data.

rjbarlow

Netstudent wrote:

I don't even see one OSPF broadcast network in that diagram. Those are all point to points.[CUT]

Yes, You are reason, that diagram was not careful, but it wanted represent a unique Ethernet LAN, and between router and router should have to be shared switches otherwise ip addresses of their interfaces would overlap.

This is the right diagram:



Maybe dtlokee was induced to his words, because he saw the same thing.

In a net like this for me there is anyhow need of at least two routers that are physically connected to router6/router7/router8 in order to have DR and BDR full sense, because, like mikearama said, if the DR fails (in this picture is intended router5 was the DR and router4 was the BDR), the BDR can't advertise routing informations coming from router6 who in the diagram is intended the unique access point to its part of the net for subnet 137.3.4.0.
My speech do not wanted to involve the problem to provide redundant physical connectivity for subnet 137.3.4.0 to all the rest of the net, this problem was not involved here, because we don't have the entire diagram of the net.

Well, thanks guys, i think I have to reflect a little more on this stuff surely.
Thanks to all.

rj

rjbarlow

dtlokee wrote:

In a multiaccess network there will be 2 routers, one designated as the DR and one desiginated as the BDR and any other routers will become DROTHER. All the routers on the multiaccess segment will establish a full adjacency with both the DR and BDR. All DROTHER routers will only establish a 2 way relationship with other DROTHER routers. So if you had an Ethernet segment with 10 routers, there will only be 17 full adjacencies with the use of a DR/BDR, vs a total of 45 if they were fully meshed(without the DR/BDR).

Dtlokee, or who feel itself to answer, can You try to explain better this concept of "2 way relationship" between two DROTHER routers? Because it can seem that it is a full relationship of the type: "I send to You and You send to me...", then it would sound against the concept of DROTHER that can send LSAs only to DR/BDR.
Thanks much.

rj

rjbarlow

OK, I have found this article at cisco.com who say:

"- Two-way: There is bi-directional communication with a neighbor. The router has seen itself in the Hello packets coming from a neighbor. At the end of this stage the DR and BDR election would have been done. At the end of the 2way stage, routers will decide whether to proceed in building an adjacency or not. The decision is based on whether one of the routers is a DR or BDR or the link is a point-to-point or a virtual link."

Practically I understand that is a step (the 3rd) in the process to form adjacency in which are a series of packet exchange in order to decide who is the DR and who the BDR. At the end of this stage only DR and BDR form adjacency to all other routers while others (DROTHERS) do not form adjacency between them. I hope not to have altered the meaning of the process.

dtlokee

The point of the 2 way relationship is to limit the amount of traffic on the multiaccess network. If all of the devices were sending the LSA's to all other routers it would needlessly increase the amount of traffic on the network.

Ok that being said, in reality all hello and link state update packets are multicasted to the neighbors so even if there was no DR/BDR you would still have efficient use of the link. Any router elected as DROTHER will send it's hello packets to the "all ospf routers" multicast address of 224.0.0.5 and all Link state update packets to the "all designated router" multicast address 224.0.0.6 which then sends them to the non DR routers using the 224.0.0.5 multicast address.

rjbarlow

Very exhaustive.
Thank You.

rj