Re: [rtcweb] draft-ietf-rtcweb-rtp-usage-12 Client-to-Mixer Audio Level

[Iñaki Baz Castillo <[e-mail]>]

2014-03-05 16:00 GMT+01:00 Sergio Garcia Murillo
<[e-mail]>:
> El 05/03/2014 13:32, Cullen Jennings (fluffy) escribió:
>
>> I am opposed to mandating that Client-to-Mixer Audio Level RTP header is
>> REQUIRED to be encrypted. It means that we can not really build conferencing
>> system where the mixers do not have access to decrypt the media.
>>
>>
> Hi Cullen,
>
> I am not sure I am following you. AFAIK DTLS is only possible to be setup
> between two peers, so in a multiconference solution the mixer will anyway
> have to establish a DTLS/SRTP session which each peer and decrypt/encrypt
> before sending to any of participants in the conference. In case of a full
> meshed conference, there is no need of a mixer at all, and the logic could
> be directly implemented on each of the endpoints.


Hi Cullen, could you please clarify this? I agree with Sergio: The
mixer (or a media-server before it) has a DTLS and SRTP session with
every participant, so it *already* has to decrypt the received SRTP
(and hence it can decrypt the encrypted RTP header as well) before
processing the media in a mixing fashion (or whichever it does with
it).

Thanks.


-- 
Iñaki Baz Castillo
<[e-mail]>