Re: [BEHAVE] IPv6 Flow Label

[Hesham Soliman <hesham@elevatemobile.com>]

Thanks for the detail Philip, I haven't read TURN for a long time.

>> => I completely agree. Philip, it might help if you explain why this
>> need
>> arises for IPv6. What is the scenario?
> 
> 
> TURN is a protocol that allows a client to "engage" the services of a
> relay (= TURN server). The driving motivation for developing TURN is
> to allow a client located behind a non-BEHAVE-compliant NAT to talk
> with other nodes. However, there are other less-common v4 NAT (and
> firewall?) scenarios where TURN is useful.  TURN is an integral part
> of the ICE/STUN/TURN suite of protocols for NAT traversal.
> 
> I realize that the question of NATs in the IPv6 world is a touchy
> subject. If there are no NATs in the v6 world, then TURN may never be
> needed. Perhaps the best answer is to wait and see what develops.
> However, like all other IETF protocols, the IESG requires us to
> consider IPv6 support when developing TURN. In same way, ICE and STUN
> also include IPv6 support. So I am trying to specify something that
> seems right for IPv6 with TURN.

=> I think that a perfectly useful answer to the IESG is to say that since
there is no _need_ for NATs in IPv6, we don't have to consider it. There are
many NAT traversal mechanisms that don't consider IPv6. There are a few that
do consider IPv6 in the context of allowing IPv4-encapsulated IPv6 traffic
to traverse NATs. But in those scenarios, I don't think the question of the
flow label or the IPv6 header in general needs to be addressed.

> 
> For the other fields in the IPv4 and IPv6 headers, TURN copies the
> field from the incoming packet to the outgoing packet except when it
> needs to do something else for relay purposes. That was my motivation
> for suggesting that the flow label also be copied (if possible). But
> from Brian's message, perhaps the more appropriate behavior would be
> to set the outgoing flow label to 0?

=> Personally, I think the appropriate behaviour is not to do anything for
IPv6 and NATs.

Hesham

> 
> - Philip
> 
> _______________________________________________
> Behave mailing list
> Behave@ietf.org
> https://www.ietf.org/mailman/listinfo/behave