Peter Hermans spent more than two decades with the Dutch telco company KPN, and is currently active as an independent consultant and program manager in several industries. This is the first part of a two-part interview.

The Early Days
Strukhoff: Before we jump into SOA, let’s talk a bit about your academic background and early career.

Hermans: Sure. I started in university in electrical engineering, with a specialization in telecommunications. I graduated in satellite communications for rural areas in developing countries. Then at KPN I started my career in the fixed line business, and was responsible for the development of network management systems. In the late 90s I became involved in the internet business of KPN, and when the “internet bubble” burst, I went over as IT director to KPN mobile in the Netherlands.

Strukhoff: So you’ve seen the entire telecom during this journey…

Hermans: Yes, and it was actually on the mobile side where I started to do some rationalization on IT. The mobile market growth was starting to flatten, so the focus shifted from “every month something new”—creating a lot of new and add-on IT—to customer satisfaction and a focus on costs. So I started architectural activities as a basis for rationalization and enabling more effective IT investments. Then the CIO said: “hey what you are doing in mobile, is actually what needs to happen for all of the company”.

Strukhoff: Hmm, good news I suppose.

Hermans: Well, yes, so I moved over to corporate in 2002 to develop the enterprise architecture for the whole KPN group. The core of this enterprise architecture was an enterprise service backbone—Gartner had not yet introduced the word SOA at that time—both in fixed and mobile. We envisioned at that time that there would be an end to the technology silos of fixed and mobile in the future. So we anticipated fixed mobile convergence. When it actually happened in 2007 we were ready for it. In IT we just interconnected the two ESBs, and the foundation for a converged company was there.

Stressful SOA, Spaghetti, and Lasagne

Strukhoff: The telecommunications industry must put a lot of stress on enterprise IT. Maybe even a leading-edge sort of stress?

Hermans: Yes, the stress comes from the fact that the Internet has a deep impact on the business model of an traditional telco. Up to now this industry revenue was earned by voice, “telephony minutes” as we call them. Having minutes migrate to voice-over-IP is really killing the revenues of a traditional telco. For example, with Skype I can make good telephony calls to South Africa for free! So telcos really have to reinvent themselves. This transformation process is underway on a global scale.

Strukhoff: Today, no matter really what vertical industry you're engaged in, people talk about SOA as if it's here and everybody's doing it. But, in fact, there's been quite a long evolution that I think is still going on. Could you trace for me a little bit sort of the evolution of SOA from its roots in Enterprise Application Integration and web services?

Hermans: I always look from the outside in, from the business viewpoint. The traditional way of interconnecting applications has always been point-to-point. But people start to realize that with that approach they create a lot of “spaghetti” that hinders the business in adapting quickly to their markets. So I think we should turn this spaghetti into lasagna.

There are two important elements to address here: processes and services. First we take process logic out of the applications. If today a business wants to modify a process, it has to make changes to a lot of the applications, which is costly and time-consuming. With SOA we shift our mindset from two-dimensional, point-to-point thinking to three-dimensional thinking--the process is in another layer. Secondly, we start to “think services,” not applications anymore, and organize ourselves accordingly.

Strukhoff: So how do we start cooking the lasagne?

Hermans: We started beyond traditional integration. Traditional EAI message-oriented solutions may be nice, but actually what you are doing is still creating virtual point-to-point connections. The first important and necessary step is to set up an integration competence center, overseeing it all.

Strukhoff: And then…

Hermans: The next step was to really introduce an ESB and start to decouple. Everybody talks about integration, but normally I tend to speak about “dis-integration,” because you want really to decouple and not to integrate! It is all about being “loosely coupled.” I think the use of web services will lead to an increased interoperability in the market, as the business wants to bundle and unbundle in a fast and flexible way in order to stay competitive. Through the concept of service orientation IT can accommodate that.

Strukhoff: Ah, very nice. A nice use of the word “disintegration.’ And I assume you mean decouple at the services level. But then what?

Hermans: I think that the evolution of web services will make it simpler, because in the end we will adapt to XML and all the standards that have emerged there. So making services available will be simpler and we will shift our focus to information, because that is where really the end user needs are.

Strukhoff: But not everything is standard just yet…

Hermans: You will always have to perform some functions as data mapping, orchestration, management and monitoring. And this is what I think ESBs, as the backbone of a SOA, should focus on in the future. This will be, so to speak, the “services” of an ESB which you can call for.

(This interview appeared originally in NOW Magazine, which retains all rights.)

Follow the author at his blog or at www.twitter.com/strukhof

• A Call-to-Action for SOA Development
• Peter Hermans Disintegrates SOA - II