Understanding the characteristics of Internet services workloads is a crucial step to improve the quality of service offered to Web users. This paper presents a hierarchical and multiple time scale approach based on [16], which propose a characterization at the session, function, and request levels. This work extends it, adding new insights to these three levels and characterizing a new level that comprises the user behavior. The approach is illustrated by presenting a characterization of a proxy-cache server from one of the biggest Brazillian federal universities. Through this case study we show clearly how to apply the methodology considering some new analysis included for the three original levels (request, function, and session) and for the new one (user). This new level of characterization is the novelty of this research area, once it considers the interaction between users and servers, that answer their requests, in order to model user behavior. This modeling shows the reaction of the users to the system performance and clarify how the quality of the offered service affects their interaction with the computer system.