Officially, integration isn't a good enough reason to do SOA. Just ask the experts.
David Linthicum says vendors are hyping the integration aspects of SOA, which is really only one small component of service-oriented architecture. Further, he thinks this puts SOA at risk, because it encourages tactical, not strategic, thinking.
Ronald Schmelzer of ZapThink contends most SOA integration projects aren't actually SOA at all. He contends most companies are really buying repackaged enterprise application integration (EAI) solutions.
More recently, enterprise architect and blogger Todd Biske seemed to support this SOA-subsumes (or should I say, consumes?) integration in a recent post. Responding to my interview with Forrester analyst Ken Vollmer, who advocated for Integration Competency Centers, Biske wrote:
"The real issue I had with some of the justifications for having an ICC was an underlying assumption that integration is a specialized discipline. While this was the case 8-10 years ago, I think we've made significant progress. I actually think there is a specific detriment that an ICC can have to an SOA effort. When an ICC exists, integration is now someone else's problem. ... It's this type of thinking that will doom an SOA effort, because everyone's first concern is themselves, not everyone else. To do SOA right, your service teams should be consumer-focused first."
All of this grumbling over integration and SOA makes you wonder: Is integration somehow bad for SOA?
I don't know the answer. But I do know this: Most of you don't care. You're building something you consider SOA, and you're doing it because you believe it will make integration easier.
Don't kill the messenger, guys, but that's what John Burke, a principal research analyst with Nemertes Research, found when he interviewed executives from 50 companies, ranging in size from 30 to 300,000 employees, about their experience with SOA. He shared some of the details of the resulting report, "Service-Oriented Architecture and Applications," with me during an interview published this week.
I asked Burke point blank what he discovered about he relationship between integration and SOA. His answer:
"We found that was the most prominent of the reasons people gave for having put a SOA in place, any kind of a SOA, whether it was meant to embrace all of their systems or just a few. The main reason that they had brought the architecture in was to integrate applications from a variety of sources."
The runner-up reasons, in case you're curious, where:
- Companies want to use SOA "as a development framework and methodology for people who build applications for themselves," and
- Because SOA comes in when they buy new tools -- whether they want it to or not.
You may find reasons to disagree with what the companies categorized as SOA, particularly the companies that said it was coming in because they acquired SOA-based tools.
But as for the number-one reason -- integration -- that's hard to rationalize away, particularly since it's not the first time a study found integration was driving SOA adoption. Remember earlier this year, when AmberPoint polled architects, operations staff and developers about SOA adoption? Seventy-five percent of companies surveyed earlier this year by AmberPoint cited integration as the "issue SOA best addresses."
It seems to be working for those companies. While most companies still couldn't quantify SOA's benefits, most companies reported it had definitely helped them solve integration and other problems, according to Burke:
"We found the majority of them felt if they had one in production that it was delivering real benefits to them in terms of reducing the cost of integration of applications, or reducing the time that it took them to bring new applications online, or otherwise reducing the cost of bringing those applications online."
In the second portion of the interview (to be published next week), Burke goes so far as to suggest integration could help companies quantify that elusive SOA ROI. But more on that later...
Like I said, I'm not defending or advocating the focus on the integration and SOA. However, with so many reporting that it works for them, I can't help but wonder, like the late great Elvis Presley, if it feels so right, how can it be wrong?