Hopefully many of you joined the 3rd International SOA symposium in Berlin. My own presentation was scheduled in the first parallel session slot. It’s titled “A Multi-Domain Modularization Approach for Achieving More Organizational Agility” and you can find the slides over here http://www.icris.nl/2010_SOA_symp_multidomain.pdf. Since the slides mainly have a graphical representation, they are a bit hard to interpret if you didn’t attend the session. Basically I told the following. First, I explained why the notion of modularity (not only for IT systems, but in general) is important for creating organizational agility. Next, I introduced a method for business modeling called DEMO (www.demo.nl) developed by Prof. Dr. ir. Jan Dietz. This method is intended to create very high level, precise business models. I’ve shown two examples on how to use these business models to achieve modularity in an organization. The first example is by defining coarse-grained IT modules by clustering automated business process steps and information objects. You can find more on this approach for coarse-grained IT modules, called BCI-3D, in this article. The second example was actor communication analysis. I used the Loomeo tool from Teseon, an ICRIS business partner, to make an analysis on the required communication between actors based on the business modules. We can use the Loomeo tool to find organizational structures with maximum cohesion and minimal coupling using dependency matrices and graphs. After we have made this analysis we can compare the results with the actual organizational structure and we can see where the mismatches and potential inefficient communication structures are.
I have to admit I didn’t see many presentations myself, always too many people to talk to at a conference like this. But one presentation I really enjoyed was the one of Steve Ross-Talbot about Savara. Savara (http://www.jboss.org/savara/) aims at making service designs (well, actually IT system design in general) testable. The Savara community says:
The aim of Testable Architecture is to reduce cost and risk in the design, implementation and testing of distributed, communicating systems. It does this by testing designs (which we might call a model of a solution) against requirements and by so doing ensure a reduction in defects later on in the software development lifecycle (SDLC). Models are created in an open-source Eclipse-based tool and are then tested against requirements.
Also, it was interesting to hear the presentation of Lonneke Dikmans about the architecture of the Municipality of Rotterdam. It seems like many SOA projects in general, and government SOA projects in particular face the same problems.
