cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
790
Views
0
Helpful
5
Replies

Is it possible to use dual hub dual cloud DMVPN in Phase 1?

Hello, I am studying DMVPN in Phase 1. I am trying to do a lab where I have 2 hubs and 2 spokes connected via 2 providers. In DMVPN phase 1, from what I understand, tunnel destination must be configured manually (gre tunnel mode is point to point). But for each spoke I have 2 hubs. How do I specify NBMA IP addresses of both hubs for the same spoke tunnel interface? I am able to specify only one tunnel destination, so only one hub.

1 Accepted Solution

Accepted Solutions

The Hubs don't need four interfaces in that case, One per ISP is enough. You end up with the following connections per Spoke:

Tun1-ISP1 <-> Tun1-ISP1-Hub1
Tun2-ISP1 <-> Tun1-ISP1-Hub2
Tun3-ISP2 <-> Tun2-ISP2-Hub1
Tun4-ISP2 <-> Tun2-ISP2-Hub2

View solution in original post

5 Replies 5

In this scenario you use two tunnel-interfaces on your spoke. Each tunnel is p2p and is pointing to one of the hubs.

One alternative solution worth trying in your studies is to use DVTIs (dynamic virtual tunnel interfaces) on the Hub instead of a mGRE interface.

Happy labbing!

I already have configured 2 p2p tunnel interfaces on my spoke routers. Each tunnel interface is connected to one ISP and DMVPN cloud. Do you mean I need 4 tunnels on spokes in this case?

Well, you could run it with four tunnels to have maximum redundancy. But that also adds much complexity. Most of the times I would run this with two interfaces, one completely on ISP1 and one on ISP2. And think about the spokes that are only single-homed. Also there you could have two tunnels, one for the termination on ISP1-HQ1 and one on ISP2-HQ2.

So, if I want maximum redundancy in this case I will have to run 4 DMVPN clouds/ 4 subnets, with 4 tunnel interfaces on each spoke and 4 tunnel interfaces on each hub? Is this true only for DMVPN Phase1? If I use Phase 2, then gre tunnels are dynamically formed on spokes and each spoke from one interface would be able to form  2 gre tunnels to both HQ1 and HQ2 in the same subnet/ DMVPN cloud?

The Hubs don't need four interfaces in that case, One per ISP is enough. You end up with the following connections per Spoke:

Tun1-ISP1 <-> Tun1-ISP1-Hub1
Tun2-ISP1 <-> Tun1-ISP1-Hub2
Tun3-ISP2 <-> Tun2-ISP2-Hub1
Tun4-ISP2 <-> Tun2-ISP2-Hub2