SIP in higher education

[Christian Schlatter <>]

Bill Reid wrote:
...
> This started out as a discussion about whether SIP or XMPP. My main 
> point was that SIP and XMPP co-exist and so that is the wrong question. 
> A better question  is whether SIMPLE or XMPP. I am leaning towards using 
> XMPP to provide messaging and presence since it is more mature than 
> SIMPLE. In reality this is not a problem since SIMPLE is not really 
> deployed.

> Here is an article talking about SIP, SIMPLE, XMPP and RTP. Granted it is 
> from Jabber.com but it seems pretty fair. Quoting from the paper
>
> "In contrast to popular perceptions, RTP, XMPP, and SIP are quite 
> complementary and provide the basis for a complete solution to real-time 
> communication needs."
>
> http://www.jabber.com/index.cgi?CONTENT_ID=619

This article suggests that SIP should be used for voice/video and XMPP
for IM&P. I see the following problems with this approach:

- SIP and XMPP use different address formats, which would introduce the
hassle of having one address for voice/video and one for IM&P

- clients would have to implement two RTC stacks, achieving
interoperability between implementations would be more difficult

- SIP and XMPP use their own authentication protocols, so there would be
the need for single sign-on (same username/pwd for SIP/XMPP)

- RTC middleware would have to support both SIP and XMPP

That is why I think that SIP and XMPP do not complement each other very 
well. I think what we will see is that the Jabber/XMPP community will 
add Jingle functionality to their clients pretty fast. And this will 
pose a strong competition for SIP/SIMPLE software clients. Especially 
since XMPP offers firewall/NAT traversal as a core functionality, at 
least for the signaling path.

As a short term solution, the SIP/SIMPLE and XMPP/Jingle clouds will 
probably get connected with gateways. But this will not be an easy task, 
mainly because they use different address, presence, and media 
description formats, and possibly also different codecs (Jingle seems to 
prefer the speex codec, which is not widely spread in the SIP community).

It is difficult to predict which of these standards will win the race in 
the long run. But since they'll offer the same features it is unlikely 
that both of them will survive. I'd like to see SIP/SIMPLE as the future 
RTC session standard since I think its design is better tailored to the 
Internet. Maybe even in combination with P2P SIP. But one should not 
underestimate the fact that XMPP/Jingle has the potential to supersede 
mass-market RTC technologies like Skype or MSN.

So I think it's important to gain experience with both of these 
competing standards, and to design the RTC middleware in a way that it 
supports multiple RTC standards. One example would be to think about 
implementing an H.350 directory that supports SIP, H.323, XMPP (not 
standardized yet) and other RTC protocols.

Christian




--
Christian Schlatter
ITS Telecommunications R&D
University of North Carolina at Chapel Hill

phone: +1-919-445-9253
sip:
h323: