Re: [sipcore] Happy Eyeballs for SIP

[[e-mail] (Dale R. Worley)]

"Asveren, Tolga" <[e-mail]> writes:
> One correction for my previous message:
> "SIP+D2U" => "SIPS+D2U"
>
> And yes, there is yet another IANA Registry to be updated as you
> mentioned and some other work, mea culpa. 
>
> There obviously will be a need to extend syntax for relevant SIP
> headers/parameters and semantics to use it as one of the valid
> protocols. OTOH, I don't think there would be thorny issues and it
> would be mostly replicating what has been done/said about other
> transport protocols.

The question of SIP over DTLS has been discussed before and the idea has
been controversial.  See, at least, draft-jennings-sip-dtls, with a
discussion starting at
http://www.ietf.org/mail-archive/web/sip/current/msg11114.html.

The major argument put in favor of SIP/DTLS is that a server can easily
handle a million clients or more.  But Roman Shpount has mentioned a
number of shortcomings of DTLS in that environment.

My personal feeling is that we'll need to devise a "DTLS Lite" that
alleviates the shortcomings of DTLS for that particular environment.
After that, the purists may still not like it, but there will be enough
practical deployment of SIP-over-DTLS-Lite to overcome their objections.

Dale