Re: [MMUSIC] Comments on ICE-TCP specification RFC 6544

[Adam Roach <[e-mail]>]

On 6/30/13 16:26, Nigel Pattinson wrote:
> Adam Roach-3 wrote
>> How many techniques from sections 5.3 and 5.4 did you implement? Did
>> your implementation make use of the following advice?
>>
>>      If the local network or host
>>      does not support IPv6 addressing,  clients SHOULD make use of
>>      other techniques, e.g., TURN-IPv6 [RFC6156], Teredo [RFC4380], or
>>      SOCKS IPv4-IPv6 gatewaying [RFC3089], for obtaining IPv6 candidates.
>>
>>
>> My experience is that Teredo-to-Teredo tunneling has success rates well
>> above 50% (at least for RFC6081 implementations) -- although this is
>> admittedly from my own personal use of Teredo, and is far from a
>> scientific survey.
> To be clear my comments about success rates are largely speculation at the
> moment. We do set up host candidates for Teredo where available. My
> understanding of Teredo is limited, I was thinking that when a
> Teredo-to-Teredo connection was established the traffic would still be
> relayed, just via a Teredo relay rather than a TURN relay.
>

I'm pretty sure that Teredo-to-Teredo traffic only has to go through a 
relay if both sides of the connection are behind symmetric NATs. It's 
been a while since I played around with it, though, so I might be 
misremembering.

/a