Visualising architectural dependencies

Abstract

Visibility of technical debt is critical. A lack thereof can lead to significant problems without adequate visibility as part of the system level decision-making processes [2]. Current approaches for analysing and monitoring architecture related debt are based on dependency analysis to detect code level violations of the software architecture [2,3,6]. However, heterogeneous environments with several systems constructed using OTS, and/or several programming languages may not offer sufficient code visibility. Other limiting factors include legal contracts, Intellectual Property Rights, and just very large systems. Secondly, the complexity of a software dependency is often greater than simple structural dependencies, including; multi-dimensional properties (as argued by [10]); behavioural dependencies [5,9]; and `implicit' dependencies (i.e., dependency inter-relatedness [11]). This paper proposes a simple modelling approach for visualising dependency relationships as an extension of the current approaches, while supporting complex dependencies. The model can be built using existing dependency analysis and general architectural knowledge; thus is better suited for heterogeneous environments. We demonstrate the proposed modelling using an exemplar, and two field case studies.