Modeling Server Workloads for Campus Email Traffic Using Recurrent Neural Networks Spyros Boukoros 1 , Anupiya Nugaliyadde 2 , Angelos Marnerides 3 , Costas Vassilakis 4 , Polychronis Koutsakis 2 , and Kok Wai Wong 2 1 Department of Computer Science, Technische Universität Darmstadt, Germany 2 School of Engineering and Information Technology, Murdoch University, Australia 3 School of Computing and Communications, Lancaster University, UK 4 Department of Informatics and Telecommunications, University of Peloponnese, Greece sboukoros@gmail.com, {a.nugaliyadde, p.koutsakis, k.wong}@mur- doch.edu.au, angelos.marnerides@lancaster.ac.uk, costas@uop.gr Abstract. As email workloads keep rising, email servers need to handle this ex- plosive growth while offering good quality of service to users. In this work, we focus on modeling the workload of the email servers of four universities (2 from Greece, 1 from the UK, 1 from Australia). We model all types of email traffic, including user and system emails, as well as spam. We initially tested some of the most popular distributions for workload characterization and used statistical tests to evaluate our findings. The significant differences in the prediction accu- racy results for the four datasets led us to investigate the use of a Recurrent Neu- ral Network (RNN) as time series modeling to model the server workload, which is a first for such a problem. Our results show that the use of RNN modeling leads in most cases to high modeling accuracy for all four campus email traffic datasets. Keywords: Email Traffic, Model Server Workload, Recurrent Neural Network, Time Series Modeling. 1 Introduction The inherently quick way of email communication, together with the ability it offers to attach files and multimedia content to messages have led to its worldwide acceptance both for personal and for corporate use. Employees tend to view emails within 6 se- conds from the time they arrive [1]. Misuse of this powerful tool is something that naturally occurs, as with every kind of technology. Irresponsible parties use its ability to carry files and/or reach numerous customers for their own, sometimes not legal, ac- tions (spam email). According to [2], Japan’s Gross Domestic Product was reduced by 0.1% due to the spam traffic. Spam emails can also break the trust in a corporation by forcing infected computers to spam as well and causing worldwide servers to block that corporation’s servers, hence isolating the corporation temporarily. Spam traffic ac- counted for 66% of the worldwide email traffic in 2013 [3]. Consequently, Internet Service Providers (ISPs), corporations and universities have to deal with millions of spam emails every day. Both spam and regular emails arrive at such great volumes that