A Model-Based Framework to Design and Debug Safe Component-Based Autonomic Systems

Building autonomic applications, which are systems that must adapt to their execution context, requires architects to calibrate and validate the adaptation rules by executing their applications in a realistic execution context. Unfortunately, existing works do not allow architects to monitor and visualize the impact of their rules, nor that they let them adjust these rules easily. This paper presents a model-based framework that enables architects to design and debug autonomic systems in an iterative and uniformed process. At design-time, architects can specify, using models, the application's structure and properties, as well as the desired adaptation rules. At debugging-time, the running application and the models coexist such that the models control the application dynamic adaptation, thanks to a control loop that reified runtime events. Each triggered adaptation is first tested at the model level to check that no application property is broken. Furthermore, architects can at any time modify the models in order to adjust the adaptation rules or even parts of the application. All changes at the model level, if checked correct, are directly propagated to the running application. Our solution is generic regarding the underlying platforms and we provide a performance evaluation of our framework implementation.