Jeroen Cloo, donderdag 08 november 2007
Businesses today face a wide range of issues that impede growth and profitability. Chief among them is the need for greater flexibility, driven by factors such as multi channel strategies, pressure to improve time to market, and the impact of mergers and de-mergers. At the same time, companies are striving for adaptive across-functional processes that can connect the silos created by ERP systems, while reducing unsupportable dependencies and costs. To overcome these challenges, business need to transform itself into a Service- Oriented Enterprise (SOE). An SOE creates business-driven value by defining and exposing its core business processes to, both the internal and external market, through the use of standardized open technology in the form of reusable services (Capgemini, 2006B). To transform an enterprise into an SOE it is necessary to identify and define those services that support in achieving the business goals. This is where Business architecture comes in.
To draw up a Business architecture, architects within Capgemini use the Integrated Architecture Framework (IAF). Modeling services call for a different mindset than modeling processes. The focus by modeling processes is to model (the sequence of) the end-to-end activities of one process and the sequence of multiple processes when processes are combined into a larger process. The focus by modeling services is to identify reusable services and to describe the output of a service. It does not focus on the way a service is realized. This shift in mindset has its impact on modeling methods: the modeling method used to identify and define services has to be able to identify services and their relation (interaction). By using a modeling method which does not fully support this (because for instance it has a different objective), the quality of the model will reduce because not all necessary elements for identifying and describing services within Business architecture of the IAF will be covered.
In this research project I developed a framework for evaluating architecture modeling methods for their suitability for modeling services and applied it to three of the most frequently used methods within Capgemini by Business architects: DEMO (Dietz, 1999) and Business modeling Method for Information planning (BMI). The framework consists of criteria for modeling methods and a classification of these criteria. A complete overview of criteria including their classification is presented in Appendix A: List of criteria.
The criteria in the framework are based on literature study, a workshop with enterprise architects (Workshop, 2006), architecture training material form Capgemini and my own experience. The framework consists of 5 groupings of criteria:
When the criteria are classified on the Seligman framework (Seligman et al., 1989) one may conclude that the way of modeling and way of working are the most important elements in the framework for modeling methods for SOE engagements. When the criteria classified as way of working are plotted on the Krogstie quality framework, it shows that the Domain Appropriateness quality aspect is the most important one followed by Comprehensibility Appropriateness.
For each modeling method two enterprise architects who applied the modeling method in SOE related engagements, were interviewed to score the method based on my framework. When looking at the scores of each modeling method one might conclude that both methods can be applied for facilitating decision making and distinguishing the artifacts of the Business Conceptual Level of IAF. BMI and DEMO should not be used for distinguishing the artifacts of the Contextual and Business Logical level of IAF. DEMO is the best method for communication purposes to stakeholder while BMI should not be applied for communication purpose.
When the total score of the modeling methods are compared one sees that DEMO has the highest score (DEMO 55.5 versus BMI 24.5 points).
When looking at the detailed score of the modeling methods, one might conclude that, even though both methods can be applied for facilitating decision making, the weak point of them is determining the impact of scenarios on all relevant aspects (people, process, management and information) and that they do not provide enough input for simulation of the system being developed. In stead of BMI, DEMO can be extended to support these artifacts. Other differences between the modeling methods are that BMI has activities to divert the business mission objectives and to obtain business goals and DEMO doesn’t and that DEMO supports an iterative way of working and BMI doesn’t.
The results of the interviews indicated also that all modeling methods do not support all criteria of the framework. Out of 35 criteria BMI does not support 19 criteria and DEMO does not support 15 criteria. From these unsupported criteria DEMO can be extended more easily. When the modeling methods are extended BMI still does not support 12 criteria and DEMO 2 criteria. Based on this information I conclude that DEMO is the best of the evaluated modeling methods for modeling services during SOE related engagements since it scores significant higher than BMI on all groupings of criteria and it can be extended easily to support more criteria.
Also the modeling method Capability Mapping was scored. Since I could only find one architects who applied this modeling method during SOE related engagements and my own experience in this modeling method is also limited, I decided not include it in the conclusion of this thesis. The method scored promising though: a total of 90 points and it can be extended easily.