Not All Networks are Equal: Empirical Analysis of Flexibility and Controllability in Software Systems
Abstract
Scholars have posited that the architecture of a system drives its lifecycle properties. For example, Moses [1] represents “generic architectures” as networks of resource flows, which he states are related to flexibility and controllability in systematic ways. This study aims to investigate these claims empirically in the context of software systems. We compare and contrast the architectures of two distinct software packages at different levels of abstraction by computing structural metrics of laterality and verticality derived from Moses’ work. Next, we investigated the relationship between these structural metrics and Moses’ proposed measures of flexibility and controllability, as well as new measures that we propose. We found that metrics for flexibility and controllability vary with Moses’ generic architectures. In the case of control flow networks, these metrics vary in the manner Moses predicts; however, when these metrics are applied to dependency networks, we found a relationship that is opposite to what was expected. This implies that measures of system lifecycle properties depend strongly on the network representation of the system used (e.g., dependency vs. control flow), thus requiring different approaches to measuring system properties.
Douglas Feitosa Tome
Alumnus
Douglas Feitosa Tome is a PhD candidate at Imperial College in London. His research interests lie at the intersection of computational neuroscience and machine learning. Current research projects explore memory formation and stabilization in the brain and continual learning in deep neural networks. Previously, he worked as an an Analyst in the Energy Advisory & Solutions practice at ICF International. He has a masters degree in Information Systems Technology from the George Washington University’s School of Business. Douglas graduated magna cum laude from the Universidade Federal do Ceará in Brazil with a Bachelor of Science in Mechanical Engineering degree. He has professional experience in financial accounting in investment banks. He has also conducted a research project focused on applying the finite element method and signal processing techniques to model acoustical sources. Within the Decision-Making and Systems Architecture his research focused on complex sociotechnical engineering systems, systems architecture, and the derivation of metrics for flexibility, complexity, and controllability of engineering systems.