In the discussion of the SECURE course 2 weeks ago, there is a topic about the routing options for DMVPN network. It mentions that If you are using full mesh mode for DMVPN, and using OSPF routing protocol, then you should use OSPF broadcast mode instead of point-to-multipoint mode for the tunnel interface.

The reason is that in DMVPN, spoke site to spoke site tunnel establishment relies on the next hop address to be the spoke site instead of the hub site, see the diagram below:

While OSPF broadcast mode satisfies this requirement, see my blog entry on OSPF broadcast and NBMA mode.

Author: Kepler Lam

Certified Cisco System Instructor since 1998

Again, I’m teaching Route course this week. Another things that students are interesting to discuss is about the compatibility of different OSPF mode in NBMA network.
I’m not going to discuss how those different mode are being used, as its been discussed in many other website. Those modes are summarized as the following slide:

If you take a look on the above summarized slide, you will find that there are 2 categories: one has DR/BDR selection (for the Broadcast and NBMA mode), while the other category does not have DR/BDR selection (other 3 modes). Actually, that is the boundary of compatibility i.e. the modes in same category are compatible.
Yet, maybe you also know that in order to form neighbors, OSPF routers will require to have consistence hello/dead interval. So even though the modes are compatible, you still need to make sure that the hello/dead interval need to be the same.
Besides, also depending on whether you are using main interface/multipoint/point-2-point subinterfaces, they have different default OSPF mode.
E.g. if you have a hub-and-spoke topology, your hub site uses a multipoint or main interface (which default mode is NBMA). Then for hub site to use OSPF point-2-multipoint mode, you need to explicitly configure. While for the spoke site, even it only has one single PVC to the hub, if you use the main interface instead of a point-2-point subinterface, you still need to define the mode in the main interface as point-2-point mode.

Following is a workable example:

Hub site:

hostname R1
!
interface Loopback0
 ip address 10.1.1.1 255.255.255.255
!
interface Serial0/0/0
 no ip address
 encapsulation frame-relay
!
interface Serial0/0/0.1 multipoint
 description Link to R2, R4
 ip address 10.1.110.1 255.255.255.0
 ip ospf network point-to-multipoint
 frame-relay map ip 10.1.110.2 112 broadcast
 frame-relay map ip 10.1.110.4 114 broadcast
!
router ospf 1
 router-id 1.1.1.1
 log-adjacency-changes
 network 10.0.0.0 0.255.255.255 area 0
!

Spoke site:
hostname R2
!
interface Loopback0
 ip address 10.2.2.2 255.255.255.255
!
interface Serial0/0/0
 ip address 10.1.110.2 255.255.255.0
 ip ospf network point-to-point
 ip ospf hello-interval 30
 encapsulation frame-relay
!
router ospf 1
 router-id 2.2.2.2
 log-adjacency-changes
 network 10.0.0.0 0.255.255.255 area 0
!

R1 and R2 can successfully form neighbors and exchange routes:

R1#sh ip ospf neighbor

Neighbor ID     Pri   State           Dead Time   Address         Interface
2.2.2.2           1   FULL/BDR        00:00:38    10.1.110.2      Serial0/0/0.1
R1#sh ip route
...
     10.0.0.0/8 is variably subnetted, 3 subnets, 2 masks
O       10.2.2.2/32 [110/65] via 10.1.110.2, 00:02:49, Serial0/0/0.1
C       10.1.1.1/32 is directly connected, Loopback0
C       10.1.110.0/24 is directly connected, Serial0/0/0.1

Please visit my other blog entry for an exmaple of broadcast and NBMA mode.

Although, OSPF broadcast and NMBA mode can also work in a partial mesh network, you need to carefully configure to make it works. There are few things that you need to cater:

1. You need to select a site with all PVCs to all other site (e.g. hut site in a hut-and-spoke topology) as the DR and properly tune the priority.
2. All sites still need to be directly reachable with each other, so if 2 sites that doesn’t have a PVC between them, then you need to explicitly use a frame-relay map to map the other sites through a PVC to a 3rd site that connects these 2 sites. E.g. spoke sites can reach each other through the hub site. (See the similar discussion for EIGRP NBMA design).
3. Sites that use frame-relay map command to reach other site cannot use the broadcast mode, must use NBMA mode and explicitly define other sites as neighbors.

Following is a workable configuration where R1 is the hub site, R2 and R4 are 2 spoke sites that each only has one single PVC to R1.

R1:

interface Serial0/0/0.1 multipoint
 description Link to R2, R4
 ip address 10.1.110.1 255.255.255.0
 ip ospf network broadcast
 ip ospf priority 10
 frame-relay map ip 10.1.110.2 112 broadcast
 frame-relay map ip 10.1.110.4 114 broadcast
!
router ospf 1
 router-id 1.1.1.1
 log-adjacency-changes
 network 10.0.0.0 0.255.255.255 area 0

R2:

interface Serial0/0/0
 ip address 10.1.110.2 255.255.255.0
 encapsulation frame-relay
 ip ospf hello-interval 10
 frame-relay map ip 10.1.110.4 121 broadcast

R4:

interface Serial0/0/0
 ip address 10.1.110.4 255.255.255.0
 encapsulation frame-relay
 ip ospf hello-interval 10
 frame-relay map ip 10.1.110.2 141 broadcast

R2 and R4 can successfully exchange routes:

R2#show ip route
...
     10.0.0.0/8 is variably subnetted, 4 subnets, 2 masks
C       10.2.2.2/32 is directly connected, Loopback0
O       10.1.1.1/32 [110/65] via 10.1.110.1, 00:03:54, Serial0/0/0
O       10.4.4.4/32 [110/65] via 10.1.110.4, 00:03:54, Serial0/0/0
C       10.1.110.0/24 is directly connected, Serial0/0/0

R4#show ip route
...
     10.0.0.0/8 is variably subnetted, 4 subnets, 2 masks
O       10.2.2.2/32 [110/782] via 10.1.110.2, 00:04:08, Serial0/0/0
O       10.1.1.1/32 [110/782] via 10.1.110.1, 00:04:08, Serial0/0/0
C       10.4.4.4/32 is directly connected, Loopback0
C       10.1.110.0/24 is directly connected, Serial0/0/0

What is the next hop for the route advertised by R4 (10.4.4.4) when received in R2? It accounts why R2 and R4 need to be reachable to each other.

See also DMVPN configuration.