Caracterização da reatividade de usuários em sistemas distribuídos

Abstract

We have been witnessing a continuous growth in the use of information technology in all areas of human knowledge in the last decades. In parallel, we may also observe an increase in the degree of interactiveness provided by computer-based systems as a consequence of either better connectivity or improved interfaces, including multimedia resources such as video, audio, or virtual reality. We may say that interactive computer systems, that is, systems with which users interact continuously, getting and providing information, become a common place in our society. From bank transactions to cell phones, we are almost always interacting with systems and even with other users through these systems. We should note that current systems allow not only human-computer interactions, but also a wide variety of user-user interactions, which are obviously richer and more complex. A significant portion of system-mediated interactions is synchronous, that is, users submit information or a request to the system and wait for a response, then submit another piece of information and the process continues. The Web, in particular the HTTP protocol, is an example of synchronous interaction. In fact, similar interaction patterns exist since the first computer systems in the 1960's. These interactions are usually complex and intriguing. It is quite hard to determine exactly the factors that led a user to behave as observed. First because the information we have about users is sparse and variable, in terms of both the instantaneous conditions surrounding the observed behavior and his or her background. It is remarkable to note that the interactions are not isolated, but successive interactions become a loop feedback mechanism, where the user behavior affects the system behavior and vice-versa. Further, we should note that the number of users in Internet services is usually large, and they interact in an indirect fashion through the server, where their simultaneous demands may generate contention. There is strong evidence that much of the user behavior is reactive, that is, the user reacts to the instantaneous conditions at the action time. The main hypothesis of this dissertation is that the real world is reactive. This concept means that the user behavior varies according to some factors related to the server and the application being provided. The main hypothesis of this dissertation is that interactive computer-based systems has a strong reactive component. The thesis of this dissertation is to research, define, characterize, and model reactivity in interactive systems, in particular Internet-based systems. Understanding users' reactivity has applicability to several scenarios, from system evaluation and simulation to system improvement, such as personalization. The main challenge is to identify what is the limit of reactivity characterization. The main motivation of this work is to understand how users interact with the system, allowing to provide services with better performance and to understand better the business applications. The contributions of this work are the formalization of reactivity concept, the specification of a multi-level reactivity model, elaboration of characterization methodologies for modeling user reactivity, and the validation of the model and methodology applying them to relevant scenarios, such as Web services and e-business. These contributions are presented in some research papers published in international conferences and journals. We foresee as benefits from this work improvements on workload characterization, new workload generation techniques that consider reactivity, novel quality of service(QoS) control techniques, and improvements on modeling and characterizing e-business services.