Richard Watson is a principal research analyst at Gartner. He has recently written about course corrections for BPM programs. He has seventeen years of experience in IT, including two years as an analyst with the Burton Group. SearchSOA.com Site Editor Jack Vaughan spoke with him.
SearchSOA.com: We did a survey recently where one interesting result was that out of the challenges the participants were facing, trying to marry BPM and SOA was the biggest. Does that resonate with what you're saying?
Richard Watson: Well, I found more people doing it poorly than doing it well.
SearchSOA.com: So it does seem to be a common challenge?
Richard Watson: Yeah, it's certainly a common thread for people who are failing. We spoke to thirty-five people from twenty-three organizations, and we didn't pick them because they were all succeeding at BPM. We picked them because they had a relationship with us and they were prepared to trust us with the information.
Very often it's the people who are failing that you learn more from than the people that are succeeding. In our case, we had a mixture, but for the people who said "We're on the second or third attempt at this thing," very often they failed because they haven't got the data modeling or the data management right.
One other thing, and this is probably the most controversial finding in the field research - I've had a fair amount of response from both vendors and users about this - I would say the industry is headed in the wrong direction in terms of primarily pursuing this seamless model-to-execution paradigm.
SearchSOA.com: It seems like model-driven architecture always looks better on paper than it actually works in real life. Is model-driven architecture making a comeback?
Richard Watson: As long as I've been in IT, we've been sort of chasing this model-driven architecture. And although the BPM tools are certainly maturing to the point where it's possible to do this, some of the people who we talked to said "Sure, we could do this, but the overhead of keeping the model and the execution environment synchronized isn't plausible."
But my point is that this is the direction that the industry seems to be heading in, and I would say that it's a mistake.
SearchSOA.com: How does the process model affect the role of the software architect?
Richard Watson: I think the software architect has the same role they've always had. If you're going to use the process model as a requirements communication vehicle then their responsibility is to make sure that the system they deliver and that the development team delivers is faithful to that model.
A lot of this is about how you use the model. And whether or not you keep the model up to date. How faithfully the model reflects the code. And I would say part of the disconnect with the model to execution paradigm is that the products create a kind of execution scaffolding. In some cases this scaffolding gets in the way of the well-known and proven design patterns that the development team uses to develop an application.
If they've got to fill out the scaffolding with their [presentation] code and their user interface code and their integration with infrastructure services like authorization and authentication, they're not free to use the design patterns that they want. They have to use the scaffolding. And that means it's going to take more effort to maintain and to deliver.
I think you'll find that that's a controversial enough point of view. It's important to emphasize that I'm not calling the models rubbish. I'm not saying to discard the models. The models are the most valuable part of BPM. I just differ in my opinion about how to use them.
Richard Watson's latest research papers include Field Research Actions to Take: Course-Correcting Your BPM Program and Field Research Results: Gaining Business Benefit from BPM.