An Architecture for a Reliability Tool Applied to Distributed Systems
摘要
Contemporary software systems, such as IoT (Internet of Things) systems, present new challenges to the development process and require new strategies for quality assurance. Reliability engineering aims to detect faults in applications, often in productive environments, providing benefits to the development activity. However, due to the inherent characteristics of these systems, complementary approaches to detect faults beyond traditional testing are needed. The Tricorder methodology evaluates resource consumption profiles as a heuristic for quality assurance, and experimental studies indicate its effectiveness in detecting new faults. This work investigates the use of the Tricorder in the context of reliability engineering to support the validation of contemporary systems, particularly IoT systems. An architecture for a reliability tool is proposed, inspired by software testing concepts. A reference architecture, RefTEST, is used to support the proposed tool. Results indicate that the Tricorder methodology can be implemented as a reliability tool, and its adoption allows for identifying anomalies that are not easily detected by traditional testing methods.