Runtime Adaptability through Automated Model Evolution

Abstract

Dynamically adaptive systems propose adaptation by means of variants that are specified in the system model at design time and allow for a fixed set of different runtime configurations. However, in a dynamic environment, unanticipated changes may result in the inability of the system to meet its quality requirements. To allow the system to react to these changes we propose a solution for automatically evolving the system model by integrating new variants and periodically validating existing ones based on updated quality parameters. To illustrate our approach we present a BPEL based framework using a service composition model to represent the system functional requirements. Our framework estimates Quality of Service (QoS) values based on information provided by our monitoring mechanism, ensuring that changes in QoS are reflected in the system model. We show how the evolved model can be used at runtime to increase the system's autonomic capabilities and delivered QoS.