A library of modeling components for adaptive queuing networks

Self-adaptive techniques have been introduced in the last few years to tackle the growing complexity of software/hardware systems, where a significant complexity factor leans on their dynamic nature that is subject to sudden (and sometime unpredictable) changes. Adaptation actions are aimed at satisfying system goals that are often related to nonfunctional properties such as performance, reliability, etc. In principle, an adaptable software/hardware system can be considered a controllable plant and, in fact, quite promising results have been recently obtained by applying control theory to adaptation problems in this domain. Goal of this paper is to provide a design support for introducing adaptation mechanisms in Queuing Network models of software/hardware systems. For this goal, we present a consolidated library of modeling components (in Modelica) representing Queuing Network elements with adaptable parameters. Adaptive Queuing Networks (AQN) can be built by properly assembling such elements. Once feedback control loop(s) are plugged into AQNs, it is possible to analyze and control (before the implementation) the system performance under changes due to external disturbances. We show the construction of an AQN example model by using our library, and we demonstrate the effectiveness of our approach through experimental results provided by the simulation of a controlled AQN.