Viptela SD-WAN design with traditional Voice implementation

[whistleblower14]

Hi all,

I recently had a discussion regarding a migration from a traditional MPLS to SD-WAN and all the nice advantages you get when using SD-WAN!

But when it comes to traditional VoIP (Signalling and Real-Time Traffic with respective codecs and qos parameters, etc.) it becomes a little bit difficult in my opinion when transitioning because in case you probably lose the benefit which an MPLS brings with the any-to-any connectivity compared to Viptela with it‘s overlay tunnels! As far as I‘ve seen there could potentially be a problem in case of when there are many locations which will need to communicate directly- and not via a Hub!

As more tunnel should be simultaneously up you‘ll need hardware power with ressources which will definitely cost money and when I go with smaller routers and maybe the possibility of on-demand tunneling I‘m not aware if this will to the job in an efficient way and without delay affecting the voice calls? so my questions is, does someone have a suggestion how such requirement can be achieved and argumented?

[Kanan Huseynli]

Hi

in case of MPLS, you also sometimes need overhead based on security and routing for direct-to-direct communication (spoke-to-spoke lets say).

In SD-WAN, you may define different VRF for VOIP or filter non-VOIP subnets and let go to Hub if it is not VOIP network, define simplified control/ data policy for site. But in any case BFD sessions between routers are needed.

Question: how many branches do you have?