Towards Model-Driven Validation of Autonomic Software Systems in Open Distributed Environments
Résumé
New distributed systems are running onto fluctuating environments (e.g. ambient or grid computing). These fluctuations must be taken into account when deploying these systems. Autonomic computing aims at realizing programs that implement self-adaptation behaviour. Unfortunately in practice, these programs are not often statically validated, and their execution can lead to emergent undesirable behaviour. In this paper, we argue that static validation is mandatory for large autonomic distributed systems. We identify two kinds of validation that are relevant and crucial when deploying such systems. These validations affect the deployment procedures of software composing a system, as well as the autonomic policies of this system. Using our Dacar model-based framework for deploying autonomic software distributed architectures, we show how we tackle the problem of static validation of autonomic distributed systems
Origine : Fichiers produits par l'(les) auteur(s)
Loading...