IMS Services Brainstorm - wrapup: IMS isn't a winner for mobile operators

I've spent the last two days as "analyst in residence" at the Telco 2.0 / IMS Services Brainstorm event in London, organised by STL. In addition to a lot of very thought provoking & senior speakers from operators, vendors, standards bodies & others, it also was an interesting physical manifestation of the telecom/mobile analyst blogosphere - especially people like myself for whom blogging isn't "the day job", but more of an outlet for disruptive opinions & publicity. Martin, James and Tomi all left their mark on delegates' thinking.

A key feature of the event was the cool wireless mini-laptop things into which delegates could anonymously type "PMIQs" - pluses, minuses, interesting comments & questions - during the sessions. Part of my job was to delve through the huge number of comments & questions and use them to help grill the speakers. The anonymity gave a lot of delegates the ability to air private thoughts that their employers wouldn't have endorsed being spoken in public, as well as the ability to criticise speakers if they gave too much marketing fluff or unsupported assertions in their pitch. (They could criticise moderators too. I was accused of being "too challenging" by one delegate, which I instead took as a compliment).

The IMS Services stream yielded some significant controversy, and, it must be said, a fair amount of negativity around some real-world practical implementations.

One thing came out clearly - IMS is not yet a viable investment for most mobile-only operators. One delegate from a very large & well-known mobile carrier described it as "a very expensive way to get simplicity" - and also mentioned that, however, they might use it to offer fixed IP telephony services over broadband (now there's a real irony.....). Unless you seriously believe that PoC or IM/presence/video-sharing usage can get big uptake & usage - and you're happy without a decent range of handsets (my usual theme), and have a rock-solid way of encouraging prepay users to forgo anonymity - then it's not justifiable. And even these services just seem to have been arbitrarily chosen because they might actually work - not because users necessarily want them. You might as well stick with SDPs and other ways of bolting together simpler services inthe shorter term - there are too many holes in IMS right now. (I'll cover the prepay issue in more depth another time - it's a very important one)

If you're a fixed operator, or a fixed/mobile combined operator, IMS offers potential benefits, if you use the concept wisely (ie don't swallow it whole). Potentially, it can reduce opex, help combine networks onto a packet core, support PSTN migration and... conceivably.... might help the introduction of some extra new services around FMC, IPTV or (yawn) videotelephony. You don't have to worry so much about device availability & the nastiness of creating decent user interfaces to make a business case (apart from FMC), as there's more hard numbers from the "plumbing" side of the implementation. However, fixed operators are much more likely to pick & choose exactly how to use the IMS functional architecture. Various speakers from BT, KPN, Global Crossing and others spoke of selecting the bits they liked, combining some elements to form "IMS Lite", and supporting parallel non-IMS domains for things like TV.

There was a large amount of apparently contradictory talk of standards and interoperability for IMS. At one level, there seems to be far too many bodies involved (variously called standards orgs, consortia, fora & various other euphemisms for "talking shop"). 3GPP, MSF, OMA, OMTP, GSMA, 3GPP2, ETSI, FMCA, NGMN, IETF and about 20 others. All of them are beavering away to create IMS interoperability "test fests", specifications, requirements, "implementation agreements" and so on. Some of them occasionally talk to each other. Some of them overlap (I got a "no comment" when I asked if the MSF's roaming/interop approach conflicts with the GSMA IPX), but despite the balloonful of hot air, many gaps are still left or exact details glossed-over. (phones, billing/charging, network security, multi-identities etc).

Bottom line on standards - too much, and yet still not enough, for IMS to be a reality.

Despite a misguided view from Telefonica ("enterprises are demanding IMS!" - really? name three, I challenge you...), there was an implicit but understated view that private SIP & VoIP is here to stay, and needs to be worked with rather than against. It's a shame there wasn't more on the IMS/enterprise interface, but I'm starting to suspect that's an even longer way out than consumer services, apart from maybe SIP-peering and things like that.

One positive trend that really came out of the event was a sense of realism. To be honest, despite continued vendor hype, there don't seem to be many operators who really believe that IMS is a panacea, or that it can exist in isolation, absorbing all the possible services & enable "defence" against the Internet. Sure, there are clearly lots of political battles within operators, and some of the more myopic "old guard" are still intent on building castles. But as one speaker pointed out, fortresses tend to encourage attacks, not discourage them. Instead, there was plenty of talk of interoperability with SDPs, broadcast and the Internet. Even Vodafone R&D had a Web 2.0 enthusiast present, and KPN mentioned P2P-SIP, mirroring another talk I'd heard from a Telecom Italia representative at another event last week. BT presented a highly user-centric presentation involving "hothousing" kids to help define new services.

In summary, I am rapidly coming to the view that IMS needs to be separated into two chunks:

- transport+ session control (plumbing-oriented, generally good for fixed/hybrid operators, if perhaps a little too complex to be implemented in one go & needing a few tweaks for things like security).

- services + applications (highly questionable, as it still needs huge amounts of work on devices, charging, ops & maintenance, and interoperability with Internet/non-IMS functions).

I'll be posting more thoughts on real-world IMS issues over the next week or so. If you missed the event and want some more info on the outputs or any other follow-up, please email me at imsforum@disruptive-analysis.com . There's going to be another similar conference in March 2007.