Using Adaptation Plans to Control the Behavior of Models@runtime

The models@runtime pattern proposes to leverage models as executable artefacts. A runtime model describing the state of the system is causally connected to the running system. Models@runtime engines typically play an active role in the definition of the adaptation plan that specifies the set of concrete tasks describing how the system should be adapted to reflect the change made on the runtime model. However, the generation of such plans generally relies on a set of predefined rules and the resulting plan is thus arbitrarily derived. In this paper we present an evolution of the models@runtime pattern with: (i) a DSL for the specification of adaptation plans and (ii) a runtime environment to enact such adaptation plans. The proposed approach has been applied to the Cloud Modelling Framework (CLOUDMF).