To be or not to be IMS, some thoughts from Madrid…

26-10-2007
IMS is really the topic of the day. By coincidence, exactly as we are preparing, in Madrid, our final oral presentation of our answers to the Telefonica IMS RFP in their brand new headquarters, we also receive by email the outcome of the second IMS RFP of Orange, a project that kept our marketing, sales and R&D busy for the past year and half.
I guess that in the future “IMS” is going to become a classic case in both marketing and engineering schools.
Marketing-wise, it is an extremely successful example of a “FUD” (Fear, Uncertainty, Doubt) strategy deployed by established telecom equipment manufacturers to fight emerging players. Years of efforts in standard bodies and lobbying passing the message ‘don’t start now, a better solution is coming’, with of course a ‘better solution’ which was tailored to their needs: a level of complexity ensuring both a healthy delay of the market allowing to continue selling existing solutions and a large price tag caused both by the number of components and integration.
Engineering wise, IMS created a dilemma, in all R&D departments, between a “R” approach (Forget all of the existing, rebuild from scratch), and a “D” approach (Start from the existing and evolve it).
The IMS standard, perhaps because it started at the peak of the telecom bubble in the years 2000 (remember the old economy was all crap ?), took a unique approach in the history of standards, it “forgot the past”. Pure “R”. My favourite example is ASN.1 BER/PER syntax: the ITU had worked for years to define what is probably the most efficient and compact data structure description tool and serialization mechanism tailored to the needs of the telecom industry. But the IMS standard used ‘text based’ with rather superficial arguments (it was “easier to parse for students”)…. Then found out that it would not work efficiently enough for mobile networks. 3GPP ended up with SIGCOMP, a stateful compressor which not only cannot be easily parsed or sniffed, but also requires the full session context to be decoded… a student nightmare compared to the good old ASN.1 .

Service providers also had a dilemma, at least those who deployed VoIP early. By the time the IMS standard was mature enough to launch a real tender, millions of pre-IMS lines had already been deployed. By the time request for tenders were awarded, and the complex integration of IMS systems had been completed, millions more lines were added to pre-IMS deployments. Even more problematic, the main reason to go for the complex IMS architecture was the way an IMS core allows for composition of services: instant messaging, push-to-talk and voice to the same endpoint and sharing the same infrastructure. By the end of 2006, it became quite clear that most of these services would never fly: push to talk and push to see were commercial disasters, SIP based IM failed and the whole market went to portal IM. Using IMS just for VoIP is a clear overkill, especially if you have to compete with service providers who are happy with silo VoIP solutions, such as ISPs or portals. R&D departments tended to push for “R” approaches (actually the best ‘book’ I know on IMS is the text of an IMS tender from a major operator, which is said to have cost millions of euros to write, and certainly was an impressive work), while sales and marketing concerns gradually increased to a point where some carriers froze their IMS projects.

All manufacturers, even those who did not believe in their own marketing, had to evolve their architecture to IMS, the FUD marketing was just too strong to resist, it would have meant suicide. The bigger groups tried both an “R” approach from scratch and a “D” approach, but in any case it didn’t translate to revenues. The typical case is to win an IMS RFP… and not deploy (or just a few hundred trial lines).

We are at this stage now… there is no question that IMS will happen, too much has been invested in the industry and I don’t see many players who would dare to explicitly stop their efforts. But it is likely to be delayed, in terms of real sizable deployments, by some years. And the biggest fear of the supplier community is that, with no business case really flying, it will be installed and sold below cost and will not cover the R&D investment… a late punishment for believing in their own marketing during 2000?

Hotel wise, the “Puerta Castila Silken” was quite a nice surprise. Not the usual boring hotel without style, everything perfect, and affordable prices.

Advertisements