Re: [Hipsec] WGLC: draft-ietf-hip-native-nat-traversal

[Miika Komu <[e-mail]>]

Hi,

On 02/27/2016 10:49 AM, Gonzalo Camarillo wrote:
> Hi Jeff,
>
> thanks for your feedback.
>
>> Regarding pros/cons:
>> How widely-deployed is STUN/TURN? Are public servers widespread?
>
> there are several of them. They are mostly used for VoIP. You can google
> for "public stun turn servers" or something similar. There are a few
> lists out there.

I guess the situation is like this:

HIP control plane relay:
* new critical infrastructure that needs to be deployed anyway (TURN 
server cannot be used for this)

Gathering of address candidates:
* from a STUN server (many available)
* ...or from control plane relay registration (which is mandatory anyway)

Data plane relay:
* using TURN server (it seems some are available)
* ...or using the ESP relay as specified in native NAT spec (none 
deployed, but I guess could co-locate with the HIP control plane relay)

So, the critical part are the HIP control plane relays which provide 
also similar functionality as STUN servers (i.e. provide server 
reflexive candidates). So I guess the question boils down to the 
availability of TURN servers.

P.S. Nothing really prevents to use STUN servers to discover address 
candidates in the native NAT traversal version. The discovery process is 
independent of the NAT penetration process.