[Standards-JIG] SIP functionality and XMPP

Ulrich Staudinger us at die-horde.de
Fri Mar 19 11:23:41 UTC 2004

Damon Lanphear wrote:

>I think it would be worth adding an "existing protocols" section to:
>that discusses how the use cases may be implemented using SIP/RTSP/SDP
>in order to highlight what may be reused, and what deficiencies are
Ok, i gathered all the existing mainstream protocols and looked at them 
critical. The updated document is available again at 
http://avtaskforce.jabberstudio.org/jep01xx.html . it was written a bit 
in a hurry and is not perfect at all, but i think it helps to get a 
quick overview and helps to judge.

My personal favorite for media is TINS/SIP+SDP+RTP and with extending 
TINS a bit to support calling through proxy (call proxy, which will then 
INVITE/call the desired partner.

The only case which aches my head a bit is the case where clients want 
to do p2p, they need to check if they can reach each other at first.
The approach then is clear,
1) check for direct connectivity
2) do tins call, either through proxy or direct
3) hand over SDP/SIP to underlaying library
4) exchange media
5) end call through TINS/SIP

imo H323 is really too ASN.1.

For conference control, a separate jep will communicate with the media 
relay/mixer component.


>On Tue, 2004-03-16 at 11:33, Ulrich Staudinger wrote:
>>Damon Lanphear wrote:
>>>I'm just getting caught up on the list discussion of IETF multimedia
>>>protocol convergence under XMPP.  It seems that there is some momentum
>>>to develop an umbrella specification that encapsulates the IETF MMUSIC
>>>suite in a "jabber friendly" protocol framework.  Frankly, I don't
>>>totally grok the scope of the current approach.  Is it that all
>>>functionality of SIP will be inherited by SIP-XML?
>>>If that is the case, then I think this is mistake.  Its already been
>>>mentioned on the list that SIP has industry acceptance and increasingly
>>>broad deployment.  More importantly, is that SIP is sufficiently generic
>>>to create a complex feature set that is difficult to represent cleanly,
>>>and implement correctly.  In other words, a LOT of work of has gone into
>>>SIP.  In this respect, re-inventing the wheel to make SIP concepts play
>>>nice in jabberland is falsely attractive.
>>>I propose that we think first about what sort of high level functional
>>>semantics we want to support.  Then we define a mapping for those
>>>semantics to the *existing* protocol suite. Finally we develop an
>>>"jabber-friendly" protocol spec that articulates this mapping.  The
>>>envisioned deployment is that existing RTSP/SIP/SDP implementations can
>>>be extended to listen for and translate XMPP-media protocol messages to
>>>RTSP/SIP requests and calls.
>>>A model just like this was employed by IETF's MEGACO. MEGACO basically
>>>provides a translation service for driving media gateways on the
>>>boundary between VoIP and conventional telephony networks. Of course,
>>>MEGACO is not of use to us other than as a protocol example of how
>>>existing communications technologies can be bridged to emerging ones.
>>>Standards-JIG mailing list
>>>Standards-JIG at jabber.org
>>To support this discussion i have already started a jep which hopefully 
>>just needs some fleshing out with xmpp commands.
>>The url is http://avtaskforce.jabberstudio.org/jep01xx.html
>>Standards-JIG mailing list
>>Standards-JIG at jabber.org
>Standards-JIG mailing list
>Standards-JIG at jabber.org

More information about the Standards mailing list