Improving Adaptive Monitoring with Incremental Runtime Model Queries

Abstract

Runtime models are often employed in different forms in self-adaptive software. They reflect, due to the causal connection, the current state of the adaptable software. Runtime model querying can be used to check whether the runtime model indicates the need for an adaptation or collect the information necessary to decide which adaptation should be performed. Given a set of runtime model queries, a natural question is how the effort to obtain and maintain the required information at runtime can be reduced. Besides the general need to reduce the overhead resulting from self-adaptation concerning its environmental impact, also restricted resources may make this a particularly relevant optimization. Two opportunities for effort reduction are the query evaluation and the necessary system state sensing. In this paper we consider both opportunities by investigating how our approach for adaptive monitoring with architecture runtime models can be improved through a better integration with an enhanced mechanism for incremental querying. We outline how incremental queries in this context can be optimized to better support adaptive monitoring. We compare different approach variants and present first very promising evaluation results that indicate that the optimized incremental queries have the potential to substantially reduce the monitoring effort and query time.