Simulating Streaming Software Applications Running on Clusters of Processors and Smartphone

Social software applications devised to process large and intensive streams of data must be usually run on complex computational infrastructure that ranges from clusters of processors to smartphones. The scalability to thousands or even millions of users is a relevant issue to be considered when designing these applications as they are not expected to collapse when they are mostly needed such as in disaster scenarios. In this context, software tools for performance evaluation of social software applications by means of discrete-event simulation have practical benefits, and yet they have not been fully developed in application domains where performance is critically dependent on massive user dynamics. This paper proposes a simulator to address this problem which combines powerful data centers and the computational power provided by mobile devices. We provide experimental evidence that shows a good agreement between actual and simulation performance measures. © Springer Science+Business Media Singapore 2016.