Triggering strategies for automatic and online service reconfiguration

The failure to meet production requirements, e.g., due to condition dynamic environment condition changes, operational performance deviations or missing opportunities to adapt, leads to a decrease of competitiveness. Technological advances are then required to promote flexible and adaptive systems with the ability to reconfigure their offered services in a cost effective manner. In spite of the current research efforts, there is still a lack of automated tools to support the service reconfiguration capability at run-time, being the understanding of when and how to reconfigure crucial to support an efficient reconfiguration process. This paper focuses on the problematic of when to reconfigure a system, with the proposed service-based multi-agent system performing the dynamic service reconfiguration based on triggering strategies. This allows to discover reconfiguration opportunities to maintain the system stable and competitive, but also to explore and promote new system configurations when needed. The envisaged triggering strategies, implemented through the intelligence mechanisms embedded in the agents' behaviors, consist in the event, periodic and trend approaches following the reactive, predictive and preventive behaviors. The preliminary experimental results validate the feasibility of such triggering strategies for service reconfiguration leading to more efficient and agile systems.