Synchronized Queueing Networks: Concepts, Examples and Evaluation Techniques

Synchronized queueing networks (SQN) are a new type of models, which allow the analysis of queueing systems subject to synchronization constraints. SQN-models originate from traditional queueing networks which have been enlarged by a new type of nodes. These nodes, called counters, can be viewed as synchronization primitives, which are used by the tasks moving through the network. Due to this combination of synchronization and queueing features SQN-models are a new approach towards the unified representation of areas which have been studied separately for a long time.The usability of this approach is demonstrated by the evaluation of models for parallel task systems and for serialization delays due to the presence of critical regions. Evaluation of SQN-models have been performed with the numerical modelling tool NUMAS, which provides various facilities for model construction and solution. In particular the use of aggregation methods, either for speeding up the convergence of iterative algorithms or for an “a priori” off-line analysis of queueing subnetworks embedded in the SQN-model, are efficient techniques to reduce the costs of an quantitative evaluation.Features for the evaluation of SQN-models have been integrated with the modelling environments of the computer manufacturers NIXDORF CAG and SIEMENS AG.