I couldn’t resist going to the session on IMS, the telecom industry’s purported salvation.
As might be expected, I’ve expressed some strong opinions on IMS previously. I’m always open to learn more and refine those opinions. Yesterday there was a good, educative session once you stripped away the slideware.
For those unfamiliar with IMS, the basic story is this: The old phone network has a media component that reaches your phone’s microphone and speaker, and a signalling part that you can’t touch (and they screw you hard when you need it to do something for you). The Internet is an IP network that just shifts bits around and doesn’t differentiate signal and media; you are in complete control. IMS is an IP network technology that re-introduces a “control plane” for signal and “user plane” for media. Bandwidth and sessions are centrally controlled and managed.
The panel was nicely constructed with analyst (IDC), vendor (Lucent, Intel) and operator (Sprint x2) views.
There’s still a lot of strangeness out there. Push-to-talk was given as a great example of an IMS application. But PTT isn’t quite real-time; there’s no QoS requirement that IMS will fix. If the radio link can’t hack it, re-arranging the packets inside an IMS box won’t make any difference.
IDC declared: “Equipment providers need to make equipment truly interoperable for IMS to be a success”. You can view this statement in one of two ways. One view says that carriers will demand a choice of vendors and low levels of lock-in. More nuanced is the possibility that services will need to inter-operate across multiple carriers. Could the mobile operators define a universal Voice2.0 application and foist it on everyone via control of distribution channels, just as with MMS? Sounds unlikely. Like SMS, Voice1.0 is a minimalist application that is good enough for a massive swathe of users. Richer apps are likely to have narrower, more targeted user bases.
IMS is a double-edged sword to carriers. On the one hand, they get a chance to compete against 3rd party applications that are eroding their revenue base. This competition doesn’t need to be ‘fair’. For example, they might only offer the connectivity fast enough for TV and videoconferencing as part of an IMS bundle, not as an Internet service. That raises the barrier to entry because it’ll be painful and expensive to build and deploy IMS apps compared to pure Internet ones. This will all feel quite reassuring to telcos, no doubt.
But the curse is that your supposedly differentiating application is now limited in scope to your connectivity customer base. If you have an application that has any form of network effect, you’ve got a problem. The Internet giants will have ten or a hundred times as many users as you. And increasingly as social networking features get integrated into IP communications, your network operator island looks rather cramped.
I’m seeing the promises of IMS as being great for feature deployment as being hollow. IMS is a Voice1.0 proposition — cheaper, but not better. A juicy qute from the floor:
What can I do with Fortran that I can’t do with Assember? Nothing. But we can write programs easier and quicker in Fortran. But the major beneficiary of IMS is the carrier, not the user.
Sprint was honest in saying IMS was enterprise-driven, a means of verticals like healthcare creating secure networks. You won’t see the Fortune 100 leading innovation in personal communications. And they won’t be held hostage to paying usurious application tolls by carriers. They’re used to buying dumb pipes, and IMS will be held to its promise of separating service and connectivity. Just a new form of mentally deficient pipe, rather than dumb pipe.
Even then, I suspect Microsoft might have a few things to say about carriers offering “enterprise instant messaging” as a service. Why stick a carrier SIP proxy between your Microsoft messaging servers? Redmondites don’t like being reintermediated.
IMS makes sense from the carrier perspective in consolidating the existing services they have into one architecture. Whether that justifies rip’n’replace on fully depreciated equipment and re-training everyone, I’m less sure. The mobile installed user base is huge, but IMS will grow up as a technology just as the access networks like Flarion, WiMax and WiFi come to obsolete the need for connectivity rationing technology like IMS.
There is the promise of seamless provisioning across multiple networks. But you have to ask yourself whether a vertically integrated, complex architecture such as IMS is really the solution. Isn’t this a fairly simple identitiy and authorisation federation problem largely solved by existing IT technology? Why not offer a simpler layered approach?
A good moderator question was what was the litmus test of whether you had ‘true IMS’. The best response, from Intel, was “if you can change you app server without changing your session server in just a week, then you have IMS.” Perhaps all IMS is about is the telecom industry discovering the difference between a web server and application server, just as the whole server business is being ripped out of their hands. A decade or two late, but never mind…
Another classic was the usual question: WHERE’S THE APPS? Specifically, what are the services a 15 year-old will want from IMS? The Ericsson response was such a classic (and representative) waffle on this that it deserves to be reproduced in full:
It’s all about their methods of communications — they drive the envelope. Everyone comes back to gaming, but I see them spending time collaborating more. We need to shorten the distance between us in a more natural and personal way. These kids would like to communicate more effecively with voice, sight, sound — as many of the senses as we can technically do. The challenge of IP multimedia is to see if we can adapt to this. Be able to replicate these solutions re-usably. Presence and availability, to see if they’re ready for a chat session, one that can give them video, voice and chat at the same time. Important thing is for them to have those choices. It’s got to be brainless, easy to use, add new value. Today they can chat, get on their video cameras on the Internet.
Err, so, remind me again … where’s the value-add of IMS? What new services does it enable?
Just to make sure, I visited the booth of IMS vendor Brooktrout. Nice demo of a game running on a PDA via IMS, but try to get out of them what IMS does for the user experience and nobody can tell me. The only user benefit again is back to that ‘seamless provisioning’. But Boingo does that today on WiFi across a zillion networks without IMS.
To add some more data points on IMS, I went to Hassan Ahmed’s IMS presentation. He is the CEO of Sonus Networks, and IMS vendor. His core message:
“It’s about empowering consumers. IM, chat, email, phone. They want to be able to seamlessly go between these services. Today’s networks don’t support that ability.”
What! I’ve been doing almost nothing but seamlessly moving between them, and folks at MSN, Yahoo!, AOL and Skype have been busy making that experience on the Internet pretty slick. There’s just no credibility to the story that IMS is fixing something that is totally broken. At best, a minor quality improvement at great expense in limited circumstances.
Hassan sees a transition from Long distance/POTS/Moble -> VoIP -> IMS; vertical integration -> converged networks -> converged services. But we can integrate services without IMS, and millions of VoIP users talk without it. QoS problems inside the edge device or customer network aren’t solved by IMS.
Why is nobody calling the bluff on this? The game is over, it’s dumbpipeville all round. A few small vertical niches with extreme security and performance needs are all that parts that require anything more.
Having see the WiFi and videoconferencing snafus the previous day, I was wondering. Would IMS have made things better. Perhaps. But the real value of IMS isn’t the technology, but the values and attitudes of telcos. IMS should really be It Mustn’t Stop. The telco attitudes to scalability, availability, and performance still retain value, even if the delivery technology doesn’t.
To wrap up, here is a grin-aloud quote from a Sprint rep:
“IMS separates signalling from bearer channel, not money from wallet.”
I think they’re on to something, there. Don’t you?
Posted by Martin Geddes at 09:18 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/567.
Thanks for this. Very helpful.
Posted by: at September 23, 2005 08:43 PMThanks for the good insight, indeed.
Posted by: at September 27, 2005 03:53 PM