I disagree that this is a "pretty good trick". Perhaps this statement just exemplifies your point that people do not fully understand IP-based services.

There are many things we can do on the Internet. VoIP is just one of them. But, one should not use H.323 or SIP for applications where they do not fit. Don't take the SIP hammer and start pounding on screws.

SIP was primarily designed for voice. It has grown since then to include more capabilities, most of which are even less interoperable than simple voice calls.

Now, let's explore the various IP services one wants. Say, streaming music. SIP? No. Wrong protocol. What about an Internet data storage system? No. Wrong protocol again. Billing system? No. Banking application? Let's hope not!

SIP is really limited in scope to providing a few basic services, primarily centered around real-time voice, video, etc.

SIP is more than 12 years old now and one should not expect significant new kinds of functionality. Sure, there are and will be enhancements to improve on the application domain for which it was designed, but it would be wrong to use it for more than that. And, I get the impression that many people do want to use it for other things– and they will likely be disappointed.

Both H.323 and SIP were developed in the mid '90s and they've reached a point of maturity. Now, users are now looking to do more and to utilize a number of different devices. After all, if you just replace a user's phone with a voice phone, what is the appeal?

For these and other reasons, the ITU started working on the next-generation multimedia system that will enable communication across a multiplicity of devices in parallel. It will use XML as the basis of the protocol and, of course, will provide a lot more application functionality than previous systems. The new project will ultimately be H.325.

Paul

Coupling of session control to nature of media:
H.323 – strong component of protocol
SIP – uncoupled by design (through SDP)

So rather than ask:
"Now, if you opened a company now, which protocol would you decide use? "

perhaps a more key question is:
If you are a network operator facing declining margins and you are continually being told that your future revenues lie with purported "IP services", but no-one can define these yet, how important is it to you to implement an access protocol which is truly payload agnostic?

Even if SIP is a 'one-trick pony' in some senses, this is a pretty good trick given the size of the financial stakes being bet against a future IP-brokered service world…

Cheers,

Micah