Framework for behavior adaptation of distributed applications

Abstract

In open environments, resource availability varies over time unpredictably. To tune resource use, adaptive systems provide the indispensable mechanisms to modify their behavior dynamically. This paper describes a decentralized framework aiming to increase the Quality of Service (QoS) of distributed applications by adapting their behavior. Based on information about applications and their execution context, the framework coordinates adaptation and finely controls applications. A model of applications exhibits the alternative behaviors that the framework selects according to the available resources. Finally, simulations show to what extent our approach improves the QoS, uses efficiently the CPU, and increases the number of completed applications. © 2010 Springer-Verlag.