Monday, February 27, 2006

All What You Ever Wanted to Know about IMS 

and even what you do NOT want to know, at least not in such detail ;-)

Martin Geddes (Telepocalypse) is discreetly hinting at the Mk2 Moriana Report on IMS, because he is providing the intro. He calls it spiffy, I would call it monstrous. If you are bound for a two weeks vacation on a lonesome island, you may take it with you. It is about 500 pages (20 MB pdf) and contains tons of manufacturers information on their take on IMS. The list of manufacturers providing answers to IMS surveys, IMS application listings, white papers, company profiles and products and services descriptions is quite impressive: Alcatel, Apertio, Capgemini, Ericsson, First Hop, High Deal, Intel, IBM, jNetX, Lucent, Nortel, Nokia, Personeta, Siemens, Tekelec and Tekvizion.

Of course I was not able to read all of it, partially because I was not so keen to read the obvious: explaining to the operators how important IMS is and why they should buy it. It is the operators which are stuck with stranded costs in some years and not the manufacturers ;-)

But I read of course the three introductory articles called IMS Analyst Opinions byMartin Geddes, Rebecca Copeland (Core Viewpoint) and by the Editor from IMS Insider.

If you take these three together, you get an excellent picture of the potential advantages and real problems with IMS.

Let's start with Martin, who is advertised in the intro by Moriana correctly as
one of truly original and independent thinkers in the field. I would recommend that readers listen carefully to his arguments and insights. They present a powerful logic and well marshaled factual evidence on the current state of the telecom industry and its likely future. Astute readers will use Martin’s vision as a pointer towards the right place to look for emerging new opportunities as well as a guide to avoiding costly errors.
Well said ;-) BTW, with costly errors the stranded costs I mentioned above are meant.

Martin starts with the clash of network cultures between IMS and the Internet by first pointing to the End-to-End principle of the Internet (the original still worth reading is here), the benefits derived from this and the consequences:
An unintended consequence of this was to eliminate the carrier as a gatekeeper function, since the meaning and value of the bits was entirely determined by the network edge nodes. The option value was increasingly redeemed by the end users as “consumer surplus” – the cumulative difference between the real value to each user and the amount they actually paid. The telecommunications industry has ever since struggled to re-invent ways of price discriminating between successful applications and re-capturing some of that consumer surplus. Many a session border controller, deep packet inspector and media gateway has been sold as a result.
He continues with:
Alert readers will already note that IMS provides security, reliability and performance guarantees that likewise involve a tension between network core and edge point functions.
I would not say provides, I would say promises, as Martin soon elaborates:
The promised benefit to the user is less spam, spit and spim; fewer dropped calls and glitchy video streams. This presumably translates into premium connectivity
prices. Furthermore, it involves some restoration of price discrimination ability, of benefit to shareholders if not all users.

IMS comes with costs, too: the opportunity cost of spending the same money on increased “best effort” capacity, plus some loss of option value in making the network smarter. The credibility of IMS stands on whether it can deliver on the following simple promise: the benefits above exceed these costs.
Can IMS deliver? He then clearly states some of the challenges, e.g.:
If a packet stream needs to traverse a single non-IMS network (or one where appropriate contracts to prioritise your packet stream aren’t in place), it’s game over for QoS promises. Most service quality issues happen inside edge devices and within LANs – where the lowest statistical smoothing is enjoyed – and well before the packets reach the service provider.
An then he comes to the point:
What about the option value part of the cost equation? For example, will heavyweight SIP scale to the smallest and weakest devices as an edge protocol? What’s the opportunity cost of fixing on SIP? After all, SIP wasn’t designed for many of the functions it is now being put to, and the resulting complexity doesn’t bode well.

If there’s one lesson to take away from the Internet protocol wars, it is that simplicity usually wins.

The same applies across the IMS technology stack. The biggest challenge is how little is needed, not how much.
Here I have the first problem, Martin. Have you ever looked at 3GPP or TISPAN IMS specifications? Take just one an example:

ETSI TS 124 228 Universal Mobile Telecommunications System (UMTS);Signalling flows for the IP multimedia call control based on Session Initiation Protocol (SIP) and Session Description Protocol (SDP);Stage 3 (715 pages)

And it is getting more and more complicated. Although some people start to talk about simple IMS - forget it. At IETF they also talked already about simple SIP - with no result.

Martin is then proposing the telocos should get in other businesses such as:
Mobile operators are already meeting success in stretching their billing platforms to non- telecommunications services such as parking meter payment. By building on web services standards like ParlayX, IMS stands a real chance of turning telcos into an application platform of choice. The ultimate irony could be that developers adopt IMS despite the complex QoS and media management, not because of it.

Perhaps telcos are belatedly discovering that the capital-intensive network is merely a loss leader in luring people into profitable data-intensive side activities.
And he is also giving some good advice, relating to the only real asset the IMS is providing to the telcos: the user identity aka SIM-card:
Why are call detail records treated like digital effluent, when they embody a treasure trove of social network data? Why can nobody but the telco build services that prevent identity spoofing? After all, it’s the telco that knows where you live, how you were credit checked, or what device you have provisioned. How come Verisign captures so much of the trusted identity profit pool? Why can’t I easily broadcast a message to everyone else in my apartment block, or neighbourhood?
The second analysis is from Rebecca Copeland, she is advertised as:
Rebecca Copeland is an IMS expert, with hands on experience in both major vendors and operators. She has a degree of technical mastery which is widely sought after and an ability to understand and analyze the market impact of new technology. Rebecca focuses on one of the key issues of the day – VoIP vs. IMS. How will this impact the industry? Who will win? Is there
room for both? Rebecca’s overview and analysis in this report, demonstrates clearly that there is no better essay on this issue in print today.
Maybe the last sentence is a bit overdoing it, but the article is really worth reading, especially the first part on VoIP over the Internet (iVoIP). I only have some problems with some of her statements in the second part on IMS, questioning her IMS expertise ;-). For example:
Just like the ISP, an IMS carrier identifies and authenticates the user before allowing any connections to be made, and the media flows directly between the conversing endpoints and does not traverse the session control servers. But unlike iVoIP, IMS media is inspected and managed by pre-determined network policy, for security and quality reasons.
Hello Rebecca, maybe you should look at some current developments in 3GPP, GSMA and TISPAN. Media flows directly between endpoints? Maybe within one network, but definitely NOT between networks. They do not even share the same IP address space. And how can IMS media be inspected, if it flows directly?

The following is also not quite true:
In IMS, it is perceived that the destination Service Providers, may deliver in-coming services to the called party as part of the session. By contrast, in a typical iVoIP connection, the assumption is that both parties are subscribers of the same service provider. So, in fact, IMS fosters cross-entities interworking, while iVoIP appear to be exclusive, making room only for one global winner.
It is true that most VoIP providers do not provide interworking with other providers, but not because of protocol reasons, but because they do not provide access via SIP URIs. But I know from my experience with IMS standardisation that cross-entities interworking is even further away in IMS yet.

She comes to this point by herself a bit later:
Although everyone agrees that IMS is the only fitting architecture for next generation communication, inertia is at work when it comes to installing it. It is true that IMS can not be justified on a one-off, self-contained service. This is the reason that early systems were equivalent to IMS-in-a-Box, all in one server. But without the IMS separation of functions, these systems can rarely scale up, or allow for multiple interworking services to be easily deployed.
The rest of the article is again well-balanced, raising some important points such as: that also the walled gardens may not be entirely safe from intrusion, that on the other hand IMS may be better suited for upcoming request by legal entities such as legal intercept and again that end-to-end QoS is a challenge for IMS too.

In her concluding remarks she states:
With the recent success of Internet VoIP, it is hard to see how iVoIP can disappear. The advent of iVoIP has changed the landscape of Communication providers, with new global players - the Internet moguls playing an important part. Their business model, with blends of communication of all types and integration with generic IT and Business systems, provides a solid basis. The spread of iVoIP has a positive effect on IMS roll out.

The threat, as well as the demand created by iVoIP, is now forcing operators to speed up their network transformation, even if it means cannibalising their own revenues. Telcos, including Cable, are not only proceeding with migration to IMS, but are also doubling their chances by offering an Internet Voice service too.

It seems that both iVoIP and IMS will grow side by side, perhaps appealing to different sectors of the market. Different users may choose convenience and reliability (IMS) at a price, rather than ‘free’ but unpredictable service (iVoIP). It is likely that iVoIP will appeal to individuals, while the security and controlled QoS of IMS will suit Business better.
I agree with everything said here, except the last statement. I am not sure if business is willing to pay more all time. They may use IMS as a last resort, but will also want to peer directly, as ECMA stated recently at TISPAN. There is also another problem: how will the local business applications within the LAN and the Internet work together with the walled garden services of IMS, without breaking the security?

And finally I want to re-emphasis: telcos and cablecos MUST NOT put all their eggs in one basket to be taken hostage by the IMS manufacturers, they must provide both services.

The last analysis is from IMS Insider and I first expected it would be somewhat biased, but to my surprise it was also quite well-balanced, but it contains OTOH some very interesting statements, e.g.:
... all the major telco representatives confirmed that ‘managing convergence’ was the key to future competitiveness. All confirmed that they are investing heavily in upgrading their networks to be able to offer IP-based ‘combinational’ services across multiple bearers and devices. And all confirmed that IMS was the key enabler to help them extract more money out of customers through improving the customer experience (through a radical shift to ‘user-centricity’), dramatically reducing ‘time to market’ (working in internet time to launch products and services) and ‘transforming the cost-base’ of a network operation.
Ok, I still have problems with "convergence", I like the statement: 'extract more mony out of the customer', and I do not see reducing dramatically the 'time to market".
A senior Vodafone representative summed up the aim: to move from “Efficient Networks to Cool Services”. To avoid becoming ‘bitpipes’. To also avoid trying to compete head-on with the internet giants. Rather, to leverage their telco skills in networking and connectivity to create a new type of service provider business (and business model) built on IP-based platforms and partnerships with multiple third parties.
They are musing themselves on this:
At present the IMS effort is firmly consigned to the network architect’s office. “Let’s get the boffins to work out how we might be able to do these new, difficult things which we’d rather not think about too much right now!” The boffins advice, of course, is “Just give us 7-10 years to do a thorough job, and we’ll create a beautiful IP-based network. In the meantime, we can trial some new service-types within, of course, carefully controlled environments.”
And then, to my surprise, they are summing up the major concerns with IMS, I could do not better:
There are a number of ‘objections’ to IMS from the commercial functions at operators today:
  • The technology’s not ready yet
  • It’ll never be fully interoperable
  • It’s a flash in the pan
  • It’s just a way for vendors to sell more kit to us
  • It’s going to be much too difficult to make it all work – we don’t have the skills.
  • There’s no real customer demand for ‘combinational’ services.
  • These ‘new services’ are nothing new. We can do them today without IMS!
  • You’ve got more important things to focus on (like growing our revenues in a saturated market)!
Also this statement is nearly true:
The competition, from all sides, is moving fast. Internet companies have 6 week release cycles compared to the 6-12 month release cycles of traditional telcos. It is the way they do business - their flexibility, their speed, their user-friendliness – which, when combined with rapid advances in technology and changes in the regulatory environment, makes them the biggest threat.
What is wrong with this? It is the 6-12 month release cycle. I never saw a telco with a release cycle such fast - it is more in the ballpark of 2 years.

To summarize: A document worth reading - if you have lots of time to spare ;-)

And do not forget: Simplicity usually wins


Comments: Post a Comment

This page is powered by Blogger. Isn't yours?