Re: [MMUSIC] A replacement for ANAT

["Christer Holmberg" <christer.holmberg at ericsson.com>]

Hi, 

>>This is what I said at the mic: we already do "replace" ICE.
> 
>Well, not quite. ICE is currently the only WG document that 
>defines how to indicate alternative transport addresses (IP 
>address and port for a particular transport protocol). Just 
>indicating alternative IP addresses is not enough - unless 
>you assume that the port is always the same.

Well, we "partly" replace ICE, then :)

>>However, even with ICE, I think one problem is that it is not very 
>>clear how to use ICE for pure offering-of-alternative-IP-addresses, 
>>when you don't use NAT traversal - or you use another 
>>mechanism for NAT traversal.
> 
>ICE does not consider this case, since it *is* the NAT traversal
mechanism. 

Well, that is the problem, if I only want to provide alternative
addresses/ports.

ANAT was not a NAT traversal mechanism, and AFAIK ICE was supposed to
replace ANAT. 

>>Let's take a simple example:
>>
>>Alice sends an offer, with a single IPv4 and a single IPv6 address, 
>>using ICE candidates. There are no STUN or TURN candidates.
>>
>>Bob only supports IPv4, so he sends back an IPv4 candidate. 
>>Bob doesn't provide any STUN or TURN candidates either.
>>
>>Now, is it is possible for Alice and Bob to start using the
>>IPv4 addresses at this point (I guess a new offer may be required to 
>>indicate which candicate was chosen), or are they required 
>>(by ICE) to initiate connectivity checks? What if the checks fail,
e.g. 
>>due to the usage of some other NAT traversal mechanism, or because
there is no 
>>end-to-end media path yet - which may be the reason that another NAT 
>>traversal mechanism in the first place.
> 
>Full ICE implementations do connectivity checks. If there is 
>"some other NAT traversal mechanism", then I guess ICE is not needed.

Well, in that case you can't offer alternative IP addresses/ports.

Regards,

Christer





> >-----Original Message-----
> >From: mmusic-bounces at ietf.org [mailto:mmusic-bounces at ietf.org] On 
> >Behalf Of Hadriel Kaplan
> >Sent: Thursday, July 30, 2009 12:15 PM
> >To: Roni Even; Simo.Veikkolainen at nokia.com; mmusic at ietf.org
> >Subject: Re: [MMUSIC] A replacement for ANAT
> >
> >
> >I'm confused by this - how does offering different *addresses* in 
> >sdp-cap not "replace" ICE, but offering different ports does replace 
> >it?
> >
> >-hadriel
> >
> >> -----Original Message-----
> >> From: Roni Even [mailto:ron.even.tlv at gmail.com]
> >> Sent: Thursday, July 30, 2009 3:43 AM
> >> To: Simo.Veikkolainen at nokia.com; Hadriel Kaplan; mmusic at ietf.org
> >> Subject: RE: [MMUSIC] A replacement for ANAT
> >> 
> >> Simo,
> >> The reason for not offering different ports in cap
> >negotiation work is
> >
> >> to not replace ICE Roni Even
> >> 
> >_______________________________________________
> >mmusic mailing list
> >mmusic at ietf.org
> >https://www.ietf.org/mailman/listinfo/mmusic
> >
> _______________________________________________
> mmusic mailing list
> mmusic at ietf.org
> https://www.ietf.org/mailman/listinfo/mmusic
>