A context-aware monitoring architecture for supporting system adaptation and reconfiguration

Abstract

Context: Modern services and applications need to react to changes in their context (e.g. location, memory consumption, number of users) to improve the user+IBk-s experience. To obtain this context, a monitoring infrastructure with adequate functionality and quality levels is required. But this monitoring infrastructure needs to react to the context as well, raising the need for context-aware monitoring tools. Objective: Provide a generic solution for context-aware monitoring able to effectively react to contextual changes. Method: We have designed CAMA, a service-oriented Context-Aware Monitoring Architecture that can be easily configured, adapted and evolved according to contextual changes. CAMA implements a decoupled architecture and manages a context domain ontology for modelling the inputs, outputs and capabilities of monitoring tools. Results: CAMA has been demonstrated in three real use cases. We have also conducted different evaluations, including an empirical study. The results of the evaluations show that (1) the overhead introduced by the architecture does not degrade the behavior of the system, except in extreme conditions+ADs (2) the use of ontologies is not an impediment for practitioners, even when they have little knowledge about this concept+ADs and (3) the reasoning capabilities of CAMA enable context-aware adaptations. CAMA is a solution useful for both researchers and practitioners. Researchers can use this architecture as a baseline for providing different extensions or implementing new approaches on top of CAMA that require context-aware monitoring. Practitioners may also use CAMA in their projects in order to manage contextual changes in an effective way.