Architecture Rationalization: From Real-World Challenges to a Methodological Solution

Architecture modeling is practiced extensively in the software industry but there is little attention paid to the traceability, verifiability and completeness of architecture designs. Deficiencies in any of these three areas in an architecture model can be costly and risky to projects. The case study reported here failed because it was deficient in all areas. We study and propose the Architecture Rationalization Method (ARM) to overcome these issues. ARM makes use of both qualitative and quantitative rationales for selecting architecture choices. Quantitative rationale uses a model based on costs, benefits and risks for choosing architecture designs. ARM also provides a method to determine when an architecture model is complete in that the level of details represented by the architecture design is sufficient.