A Multi-dimensional Component Importance Analysis Mechanism for Component Based Systems

In a component-based system, both the reliability and topological location of each component contribute to the software architecture reliability. Due to the limited resources, it is very crucial to identify the key set of components such that we will be able to devote our resources to those components in order to improve the system availability and reliability most effectively. Unfortunately, few works have been dedicated to the analysis of component importance and the system reliability from the aspect of components´ location. This paper introduces a multi-dimensional mechanism to analyse the importance of components in a system and the their influences on the reliability from four aspects: centrality, entropy, costumer preference and service priority. This mechanism provides an interactive approach to identify the key set of components during the architecture design phase. Both centrality and entropy evaluate the component influence from static perspective. On the other hand, we estimate component importance from the dynamic point of view with costumer preference and service priority during application execution.