Predictive Analysis of Cloud Systems
Patrícia Araújo de Oliveira · 2017
Predictive analysis methods offer the possibility ofestimating the impact of design decisions, which may help inthe accomplishment of operational optimal results, before thedeployment of the system, and therefore minimizing the requiredeffort and cost. However, current predictive methods cannot beused on cloud environments, because of their complexity anddynamic nature. The main goal of this thesis is to investigatemethods for predictive analysis of cloud systems. Given modelsof cloud systems and their environments, we will specify differentadaptation mechanisms, and techniques for the estimation ofdifferent QoS metrics that help in the analysis of cloud systems. Specifically, we will use model transformation techniques tospecify the behavior of systems and their dynamic adaptation, as well as the performance metrics and tools for their analysis. Our tools will be based on simulation, and we will explore theuse of statistical model checking tools, and in particular thosedeveloped for graph-transformation systems.