Towards Cloud Storage Services Characterization

Monitoring of Internet services shows that there is a global and growing trend in the use of Cloud Services. This paper aims to identify and quantify the use of Cloud Services taking the University of Minho (UMinho) network as a practical case study. Thus, this study focuses on characterizing Cloud Storage services, identifying the most accessed Cloud Storage Providers and the characteristics of corresponding traffic. As a first step, this involves identifying appropriate techniques for traffic classification and the definition of a model for processing the collected traces. Cloud Storage services present several characteristics that turn the current classification methods insufficient or too complex to apply, namely the use of dynamic communication ports and security protocols encrypting the traffic. This has motivated the use of a new classification approach based on Tstat tool, which allows extracting signatures of servers during SSL handshaking. The obtained results provide global statistics regarding the most used services at UMinho, focusing subsequently on Cloud Storage services. For these, the top Cloud Storage Providers within user preferences are identified and the corresponding traffic characteristics discussed.