Abstract : With the development of the communication infrastructures, the number of applications collaborating at large scale increases. To maintain, and continue to deliver services of good quality to the end-users, very-large-scale applications continuously adapt themselves, depending on the changes in their surrounding. Stabilization of the system thus becomes a keystone issue in the adaptation process, in order to reduce the system reconfiguration cost. Existing approaches, for the stabilization of very-large-scale systems, provide solutions that are partially efficient. For example, learning-based stabilization algorithms give good results in predicting application behaviors, but still suffer from their weak reactivity. In this paper, we propose an approach of combining different goals-oriented stabilization algorithms, in order to provide sustainable and efficient stabilization for large-scale systems.
