How Practical is your SDP?

Members of the Telco 2.0 team went on a trip to the Italian Lakes this week, where we were stimulating and facilitating an impressively organised conference for Service Delivery Platform experts jNetX and a gaggle of telco people from most parts of Europe. We used a basic version of our interactive Mindshare approach to elicit audience feedback on the issues raised.

Some people there had a funny reaction to our use of the word "platform", central as it is to the Telco 2.0 vision - isn't a platform really an operating system? Or something used when drilling for oil? They're right, of course; a Telco 2.0 platform is a sort of operating system, just as Salesforce.com or Amazon's IT infrastructure can be seen as a sort of single huge computer. But it became increasingly clear that this cuts both ways; you need the right platform in the second sense to do a platform in the first sense right.

Where did it all go wrong, Telco 2.0?

jNetX's actual product illustrates this - it's nothing if not complex, being a mighty stack of JavaEE elements that deal with SS7, IMS, and pure Internet access networks at one end and generic IT (and Web 2.0) interfaces at the other. What is noticeable about it, though, is that it is made up exclusively of stuff that already has Java Specification Request (JSR) status (especially JSRs289 and 240); it's all open-standard if not open-source.

And the examples of its use in practice bear this out. At BT, head of tech strategy Mark Kent described its deployment for new IN services at BT Global, with the assistance of a tiny startup, NetDev. Crucially, the role of systems integrator remained with BT's chosen partner, and the standard interfaces meant that BT remained free to choose whatever hardware, software, and other partners it liked; standards compliance is the death blow for vendor lock-in.

Hence BT's spanking new apps platform, Project Firebird; running in a JAIN SLEE environment, it acts as a "resource adaptor", speaking assorted Java APIs to IT systems in one direction, SIP to switches and routers in another, and the Linux API to the carrier-grade Linux machine it runs on. As Kent says, "you need to look out for platforms." In every sense; Telco 2.0 is well aware of the dangers of the oil-drilling kind.

At Mobilkom Austria (with operations in lots of emerging markets), Alexander Kuchar described the process of developing their group-wide "mobile service network", a term which describes everything between the RAN and the IT system. Kuchar decided to go for a full-blown IMS, lock, stock, and ten wriggling tentacles, and use jNetX's Telecom Application System as the IMS applications layer. The upshot? The second time around, launching a VPN service cost just 10% what it had the first time. Interestingly, he said that the most important factors in successful deployment had been the standard nature of the system and the fact that Mobilkom retained full control of intellectual property throughout, rather than the specific IMS architecture.

Telco 2.0's scepticism about IMS is well-known; but what is most Telco 2.0 about all this is jNetX's DNA, not just a rather strained biological metaphor but a Developer Network Area, which brings together its community of third-party Java developers. JNetX markets their applications to its telco customers, making itself probably the only vendor to have a catalogue of 100 mobile apps. As they point out, there is no killer app; but that's a good thing. All the better to develop as many small and good ideas as possible.

If we had a criticism, it would be that perhaps some denial persists about the "special nature of telco hardware" - quite a few people claimed that such-and-such a service would never scale on IT gear, even though we heard a nice anecdote about an unnamed telco boss who demanded to know what platform (that again) a new apps server runs on. "Linux, on an IBM Bladecenter," our man responded. "So you're just buying a computer?," the boss said.

"Well what do you think our network runs on?" Indeed.

Further, Telco 2.0 didn't think there was quite enough interest in external innovation - getting out of the operator's development shop. Fortunately, there was jNetX boss Gary Miles on hand ("Operators don't do mashups!"), and there was this thought from Huawei Europe CTO David Fowler.
"We can either throw money at our own development, or partner with others so as to start running today.."

Fowler also believes there is no consumer demand for mashups, and therefore vendors should concentrate on supporting operators' cost saving efforts, for the next three years at least; the obvious answer to this is "if there is no demand, why do so many people on the Internet spend their time making them?" It's central to the Telco 2.0 vision that cost-cutting is not enough; an industry whose primary problem is that its basic product is getting progressively cheaper needs new sources of revenue.

We finished the event by asking the eminent panelists what they thought were the "big things that need doing to stimulate growth for the mobile industry"; below were the responses. We think they bear this point out.

  • FUNCTIONAL SEPARATION IN MOBILE AS BT HAS DONE IN FIXED
  • BELIEVE IN THE ART OF THE POSSIBLE AND EXPERIMENT
  • CHANGE PROCESSES INSIDE OPERATOR ORGANISATIONS TO BE MUCH MORE FLEXIBLE AND RESPONSIVE
  • MUCH FASTER TIME-TO-MARKET; LEARN TO "RELEASE EARLY, RELEASE OFTEN"
  • COURAGE TO CHANGE OUR WAY OF THINKING IN AN INCREASINGLY 'OPEN' WORLD.