[VOIPSEC] Why a secure keyechange for media encryption?
Hadriel Kaplan
HKaplan at acmepacket.com
Fri Apr 28 15:05:17 CDT 2006
Hey Mark,
Comments inline...
> -----Original Message-----
> From: Voipsec-bounces at voipsa.org [mailto:Voipsec-bounces at voipsa.org] On
> Behalf Of Mark Baugher
>
> On Apr 28, 2006, at 11:23 AM, Hadriel Kaplan wrote:
>
> > Depends on how and through what.
>
> I assume you're referring to S/MIME.
Yup, just S/MIME.
> > Service providers use many boxes that
> > either need to see inside the SDP, or need to change it.
>
> Certainly you can see inside the SDP and use S/MIME (multipart/
> signed). And an intermediate system can change it, as far as I
> know. (I have read the major S/MIME-related specs and toyed with it
> in the OpenSSL library but I'm not a practiced expert.) It might be
> tricky, but why couldn't someone duplicate a part, sign it, and send
> on both parts? This likely requires a software upgrade to SIP box.
For just signed you could see it, but if you change it and re-sign, it
wouldn't be signed by the right party, would it? I think an SBC vendor
tried that once and it failed miserably, or maybe they just did. :)
Of course doing it ain't that much fun in terms of scalability anyway.
And our problem is bigger than just one B2BUA box, it's a chain of them. A
service provider has little control over what vendor/equipment their peering
partners wish to use. That's one of the reasons I think the sip-identity
draft is going to have issues. It's most useful case IMHO is crossing a
chain of providers, but since it signed the SDP (and contact) it'll break
most of the time. I like the idea of it, but it decided SDP was part of a
user's identity.
> > SBCs, media
> > servers, transcoders, etc., often change it (though they don't have
> > to in
> > all cases).
>
> Yes and they need the key also. If mixers, transcoders, SBC's etc.
> terminates one RFC 3711 crypto context and originates another, then
> they need to be involved in the key establishment; in other words,
> they need to be trusted to have the keys.
SBCs don't typically need the SRTP keys, but sure if they're also
transcoding or doing dtmf inspection or are providing the SRTP termination
then they would. And if they support TLS and sdesc they'll get them. Just
like a PSTN gateway would need to.
> > Lastly, hardly any phones or gateways support it, so you won't get
> > much
> > success for your trouble.
>
> Are you saying that most SIP devices choose not to support MIME as
> well? Or is this strictly an S/MIME issue?
Of course just talking about S/MIME. :)
-hadriel
More information about the Voipsec
mailing list