Web applications usually involve a number of different software libraries and tools (usually referred to as frameworks) each carrying out specific task/s and generating the corresponding overhead. In this paper, we show how to evaluate and even find out several configuration performance characteristics by using virtualized environments which are now used in data centers and cloud environments. We use specific and simple web software architectures as proof of concept, and explain several experiments that show performance issues not always expected from a conceptual point of view. We also explain that adding software libraries and tools also generate performance analysis complexities. We also shown that as an application is shown to scale, the problems to identify performance details and bottlenecks also scale, and the performance analysis also requires deeper levels of details.Instituto de Investigación en Informática