Telepocalypse

OPINION://Teh wrogn kiind f niens

I’ve been reading more on IMS and its apologists.

In a nutshell, IMS is an open architecture designed for the convergence of data, real-time media, and networks. IMS can encompass all the wireline and wireless networks we employ at home, in the office or on the move. In other words, it enables the creation of converged, unified domains.

That’s obviously not the meaning of “open” you would colloquially use: the technology is in principle open, the deployment most certainly isn’t. In tribute to artist René Magritte, ceci n’est pas une pipe: IMS isn’t an open standard, just an illusion of one.

Errr, and last time I checked the thing that encompassed all the networks was called the Internet, and the real-time media and networks were diverging at a rapid rate of knots. IMS is not a very compelling illusion, I have to say.

If your retirement dollars are invested in a telco, you should be urgently asking yourself what this stuff does that Skype doesn’t do — and Skype’s only a whiff of what’s possible.

Ah, here it is:

These are historic IP comms issues. Traffic on the Internet is subject to delay and jitter but there are various QoS workaround mechanisms and solutions. The IMS, however, addresses the problem head on.

During a SIP session the user’s device negotiates its capabilities and expresses its QoS requirements. When this process is finished, the device reserves suitable resources from the access network and when end-to-end QoS is established, the packets are encoded with the relevant protocol (e.g. RTP).

So I guess that conversation I had with my brother las night while he was downloading a ton of BitTorrent stuff and my wife was playing a networked game on our PC was just an illusion, too. Yeah, the voice quality dipped to good cell-phone levels from Skype’s best-quality FM-like offerings. But who cares?

IMS’s sole justification is quality of service bonded with application intelligence. Take that away and the house of cards falls down.

There’s more:

There is a caveat: operators must have Service Level Agreements (SLAs) in order to guarantee the QoS in the interconnecting backbone.

Oops! So if your IMS stream traverses a single packet network that doesn’t participate in some pointless bandwidth-reservation scheme, it’s over. Hope your internet cafe, hotel, school and coffee shop are all wired up to the scheme. Don’t forget to tell your mum to upgrade to an IMS provider, too, so everything works smoothly when you’re visiting.

And heaven help you if your requirements change mid-session or you don’t have a constant bitrate application, ‘cos you’re going to be sending a lot of signalling messages up and down trying to renegotiate paths all the time. ATM to the home failed for good reasons, folks. Sprint ION was the billion-dollar experiment that showed us that road is closed. IMS is ATM mk 2 — now failing at a screen near you. Different technology, same business model.

Even more important is: what’s the point of reliability and quality of service? What I want is successful communications. This is not the same as reliable single links in the chain. David Isenberg once wrote about buying as many nines as you need by bonding together different access networks. This is true, but my argument is you additionally need to look at the application layer to see how value is created. Connectivity alone isn’t end-user value. Value is created by sequences of interactions that meet a user need at the least effort and expense. (Or to put it another way, the option value of the stupid connectivity has to be redeemed by an application at some point to make the option worth having.)

In traditional PSTN and mobile networks, you spend a lot of time manually working around the limitations: no presence; strange numbering schemes that confuse people, places, devices and services; awful UIs to messaging systems; lack of directory integration, and so on.

The result is voicemail ping-pong, unresolved converations, unheld conference calls, missed messages, forgotten replies, pointless calls to your secretary to get contact numbers, inappropriate interruptions, and so on.

Forget five nines. From the perspective of the user, voice telephony doesn’t yet deliver one nine of success. Do you really believe over 90% of the calls you place meet their “objective” first time?

Now some might argue that IMS will deliver on all this too. Bollocks, I say. We don’t know how advanced forms of presence and new UIs should be blended together. The MIT Media lab is still working on it. IMS requires all operators everywhere to co-ordinate on deploying new features if they’re going to be useful to users. What’s the point of a feature that only people who bought connectivity from the same source can access? It’s like having a supermarket only catering to drivers of Toyota cars.

Skype is barely any better. You can use a common address book to IM me and see if I’m good to chat; and you can see I’m online and you have a chance of not getting voicemail. But what’s with just the red and green answer/decline buttons? Where’s the amber “will call you back later” button? The “screen to voicemail” button? I believe we can deliver these things without making the UI too complex for new users. But if Skype screws up, someone else can quickly take its place. IMS doesn’t admit that possibility; it’s stuck in the “telco as feudal overlord” mode of thinking.

Telcos as gatekeepers to new functionality doesn’t exactly have a great track record. Expect the commercial damage that is IMS to be routed around. The idea that “surely they must know what they’re doing when spending all these billions” just dosn’t hold any more. Telecom has destroyed enough shareholder value via bloated spectrum auctions, overpriced take-overs and outright accounting fraud. Don’t feed the monster with you own money, because IMS is the hope you’re betting on. And IMS is fucked.

UPDATE 1: Just in case anyone gets the wrong end of the stick, Bob’s got a great new blog and I read all his stuff, as well as the other VON Magazine colmnists. I’m just using his text to illustrate the standard IMS story that’s being spun. Bob knows the real deal.

UPDATE 2: IMS will, in my opinion, still be deployed. But it won’t create any premium service revenues of merit or any pricing power. The price of “voice service” within trust domains (among friends, same company, etc.) will be zero. Just a very expensive inflexible infrastructure which requires an army of people to negotiate and monitor interconnect agreements which are an order of magnitude more complex than today’s peering/routing. More cost. Less revenue. Profit!

UPDATE 3: Is anyone else a bit suspicious of the claims from BT that their 21st Century Network will save a bucket of cash? Isn’t the capex already fully depreciated on their current switches? Can’t they do a phased replacement and cannibalise spares bit by bit when vendor support ends? Is their electricity bill for old iron switches really that big? Can’t they automate away any of the management costs? Won’t they have to retrain a ton of people to manage an IMS network? Why not let the workforce age out along with the obsolete equipment? I don’t get it — why the rush to IMS? Genuine question, answers please!