OSPF Point-to-Multipoint Non-Broadcast Network Type over Frame-Relay

Lessons Discussion
22

so I recreated this topology in GNS3 to see WireShark capture of OSPF messages, I could not get it to work because you did not specify “no shutdown” commands for s0/0 in the configuration section of your post.

1 Like
23

Hi Rene,

On Multi-point Frame relay is there a reason that a single subnet is used over multiple subnet sub-interfaces. The typical Frame relay point-to-point it seems that its a defacto standard to use different subnets with sub-interfaces and with multi-point its typical that the same subnet is used and no sub-interfaces. WHat is the reasoning for this standard. It seems to me, that either method could be used for multi-point both sub-interfaces(different subnets) and single subnets (no sub-interfaces). I am missing why one method is used or preferred over the other???

Thanks
Mike

24

I was unable to get it to work in GNS3 as well because of Frame Relay Switch will not configure in a mesh configuration. But if the same setup is entered into Packet Tracer it works perfectly without any issues. Konrad a suggestion, by default OSPF will not start without at least 1 interface being up. That is why it is always good to configure interface loopback 0 prior to configuring the Serial interfaces. Then when you do your router ospf <area> command, OSPF will run irregardless of your interface being shutdown. I am using virtual labs as well, there are small differences between my Cisco IOS V12.4 and the latest version 15 in Packet Tracer latest release. You can get Packet Tracer from CISCO at no charge, UDEMY Instructor Lazario (Laz) only uses Packet Tracer in his videos for CCNA and CCNP. So, get Packet Tracer if you can, it has the latest Cisco IOS already installed on routers and Switches.Latest Cisco IOS V15 the ip ospf network point-to-multipoint has been deprecated and replaced with ip ospf network broadcast command. FYI

1 Like
25

Hi Mike,

The idea behind frame-relay multipoint is that you can use a single subnet for all your PVCs. It doesn’t matter if you use the physical interfaces or sub-interfaces for this. By default, a physical frame-relay interface is multipoint.

As a best practice, it’s best to always use sub-interfaces. Let’s say you have one router that is connected to 1x point-to-point network and 2x point-to-multipoint networks. It would be best to use 3 different sub-interfaces for this, it’s a bit cleaner compared to configuring one on the physical interface and creating two sub-interfaces for the other.

With point-to-point, each PVC uses a different subnet. Since the physical interface is multipoint by default, you should use a sub-interface for this.

The downside of using multipoint is when you use it on a hub-and-spoke topology. Your routers think there is full mesh reachability, while in reality, everything goes through the hub. Distance vector routing protocols like RIP/EIGRP have issues with split horizon.

OSPF works, but requires some specific configuration depending on the OSPF network type you used.

The OSPF point-to-multipoint network type is not deprecated btw, it’s still there and behaves differently than the broadcast network type.

Router(config-if)#ip ospf network ?
  broadcast            Specify OSPF broadcast multi-access network
  non-broadcast        Specify OSPF NBMA network
  point-to-multipoint  Specify OSPF point-to-multipoint network
  point-to-point       Specify OSPF point-to-point network

Router#show version | include IOS
Cisco IOS Software, IOSv Software (VIOS-ADVENTERPRISEK9-M), Version 15.6(2)T, RELEASE SOFTWARE (fc2)

Rene

1 Like
26

19 posts were merged into an existing topic: OSPF Point-to-Multipoint Non-Broadcast Network Type over Frame-Relay

27

Hey Rene/Laz

Can you please explain a little bit more? Why don’ t we have to configure the neighbor command on spokes? and the logic behind it?

Because I am labbing for EIGRP static neighbors and this is not the case.

28

Hello sales2161

When you have a Point-to-Multipoint Non-Broadcast network type over Frame-Relay and you want to configure EIGRP, no automatic neighbor discovery occurs. Th is because it is a non-broadcast network type. For this reason, neighbors must be configured manually.

Now in order to configure the neighbours manually, it is only necessary to employ the neighbor command on the Hub. Once this is done, the spoke routers will receive the EIGRP packet being sent to them and will respond to it creating the adjacency.

I hope this has been helpful!

Laz

1 Like
29

Are we assuming that lmi is auto generating dlcis? Can the spokes communicate with each other without frame relay maps?

30

Hello Justin

With the configuration as is in the lesson, the DLCIs are determined using LMI. Inverse ARP is enabled by default so hub to spoke communication is enabled. You can use the show frame-relay lmi command to see the DLCIs issued and the LMI information.

However, you will find that spoke to spoke communication is not possible. Because the spoke routers do not have direct connectivity with each other, dynamic Inverse ARP would not work between them. Dynamic Inverse ARP relies on the presence of a direct point-to-point connection between two ends. In this case, dynamic Inverse ARP only works between hub and spoke, and the spokes require static mapping to provide reachability to each other.

You can find out more information about Frame-relay and how to configure it at the following lesson:

I hope this has been helpful!

Laz

1 Like
31

Hi dear team.

  1. What are the possible combinations of OSPF Network types? I mean, if on one router interface I use non-broadcast ospf network type I can use non-broadcast ospf network type on another router but what more combinations?

  2. What are the tipical cases I can use theses Ospf Network Types?

  3. What to say regarding EIGRP and Frame-Relay ?

32

Hello Zacarias

You must use the same network type on OSPF routers that become neighbors. This is because each network type changes the way that neighbors are created, and how (and if) DR/BDRs are elected.

For typical cases where the various network types are used, take a look at this post which should answer most of your questions.

Finally, EIGRP can be configured for frame relay, with some special configurations. Take a look at this lesson to find out about these scenarios.

I hope this has been helpful!

Laz

1 Like
34

@lagapidis Do we need to configure network command to activate EIGRP on interfaces? like OSPF does in this frame-relay topology?

35

Hello Nihar

Yes, the network command is necessary to allow an EIGRP router to know which interfaces are going to participate in the routing protocol. If an interface has an IP address that is within the same subnet as the address configured on the network command, then two things happen:

  1. The network of the interface begins to be advertised to EIGRP neighbors
  2. The router can establish EIGRP neighbors through that interface

So if the network command is not used on a particular router in this topology, then that router will not be able to establish any EIGRP neighbor adjacencies with any other EIGRP routers.

I hope this has been helpful!

Laz

and their subnets are going to participate in the routing protocol. This includes both advertising If the network command is not used, then no local networks are being advertised

1 Like
36

Hi @ReneMolenaar

I found your explanation pretty useful. If my understanding is correct, the Hello and LSA packets are sent in the following manner for the different Network Interface Types:

  1. Broadcast Network: Supports Multicast and Broadcast, so both hello and LSAs are sent using the multicast address. Example: Ethernet.

  2. Non-Broadcast Networks: According to RFC 2328.
    a. Default: NBMA - No broadcast support but OSPF neighbor discovery and adjacency are built and maintained using multicast addresses by default. Since multicast/broadcast is not supported, hellos and LSAs are sent as unicast messages. Also, the neigbors need to be configured manually since this is a multi-access network. Finally, OSPF will need to be activated by using the default “network” command.
    b. Point-to-Multipoint: Essentially the OSPF router sees a bunch of point-to-point links.
    So in this case, since it doesn’t support broadcast, hellos and LSAs are sent using the neighbor unicast address. No DR/BDR election since it is a p2p link.

  3. Point-to-Point Networks: Typically 2 routers connected via a Serial Cable that run PPP.
    Here there is no DR/BDR election since each router knows there is exactly one neighbor at the end of the cable. However LSAs and Hellos are sent via Multicast addresses instead of Unicast even though it is a point-to-point link.

Questions:

  1. Why is there a disparity between the way hellos and LSAs are sent in a point-to-point and point-to-multipoint link? Though Point-to-point links typically have only one router at the end of the link, why can’t we use Unicast instead of multicast? Why does OSPF chose to use multicast?

  2. Can we have broadcast frames over a Serial link? Since it uses PPP, the l2 frame structure is different from Ethernet. Reason for this question is to understand why multicast is used on a p2p link for hellos and LSAs?

37

Hello Adhithya

It looks like you have a good grasp of OSPF and the various manners in which the Hello and LSA packets are sent on different types of networks. Let’s address your questions:

The reason OSPF uses multicast even on point-to-point links is because of efficiency and simplicity. Multicast allows the router to send a single hello or LSA packet that can be received by all OSPF routers on the network, instead of having to send individual unicast packets to each router. Even though there’s typically only one other router on a point-to-point link, the multicast method is still used because it’s the default behavior of OSPF and it simplifies the protocol’s operation. If unicast were used, the IP address of the other end would need to be known (i.e. configured manually on each end), and that requirement also eliminates the automatic nature of the OSPF neighbor adjacency establishment.

As for your second question, it has to do with the mechanisms operating at each layer.
At Layer 2 of a serial link, there is no concept of multicast. Indeed, I believe that many Serial technologies such as PPP for example, don’t actually include a Layer 2 address in the framing because there is only a single host expected on the other end of the link, so addressing really has no meaning. So at layer 2, there is no multicast addressing and no concept of multicast. However, if you are running IP at layer 3 on top of such a serial link, you can use multicast IP addresses. They will still be sent over the link as a point-to-point communication at Layer 2, but the multicast addressing will still work. So since OSPF uses multicast addressing as answered in your first question, it is also used here.

I hope this has been helpful!

Laz

1 Like
38

@lagapidis Thank You so much. I was almost on the verge of being overwhelmed by the concepts of OSPF interface, Network and Area Types. The articles on your website have been fantastic, and the forum is equally good. I am eagerly awaiting your answers to other sets of questions on the LSA types :grinning:

Regards,
Adi