Hello David
DMVPN does indeed use GRE as its underlying tunnel-creating mechanism, however, it also uses NHRP in order to dynamically resolve the next-hop router’s IP address. These commands are applied for the benefit of the operation of NHRP.
On the hub, we issue the ip nhrp map multicast dynamic command to tell the router to dynamically learn and map multicast sources to remote sites. When the hub receives multicast traffic, it will use NHRP to determine which remote sites are interested in the multicast group and then forward the multicast traffic only to those interested sites. If the command is not issued, the hub router will treat multicast traffic as unicast traffic, sending a copy to each remote site regardless of whether they have active receivers interested in the multicast group.
On the spoke, we issue the ip nhrp map multicast <ip_address> command where <ip_address> is that of the hub. This command essentially defines a mapping for multicast traffic, so that the spoke will send all multicast traffic to the hub. If the command is not issued, the spoke may send the multicast traffic to multiple hubs (if they exist) or it may even broadcast it over the entire network.
These commands essentially apply mappings that tell NHRP how to handle multicast traffic in a more efficient and correct manner.
I can see how these commands may seem somewhat redundant, however, they serve different purposes.
The ip nhrp map command maps the hub’s tunnel IP to its public IP. This defines the map[ping between the logical tunnel and the physical NBMA address of the hub. This is needed to perform address resolution to help spokes determe where to send NHRP requests.
The ip nhrp nhs command specifies the IP address of the NHRP server for NHRP registration of the spokes. The key here is that although it’s common for the hub router to also serve as the NHRP server, this is not always the case. In larger and more complex DMVPN deployments, you may choose to have a dedicated device other than the hub serve as the NHRP server. In that case, the IP address of the server and the IP address of the hub’s tunnel IP will be different.
I hope this has been helpful!
Laz