Evolutionary visual software analytics is a specialization of visual analytics. It is aimed at supporting software maintenance processes by aiding the understanding and comprehension of software evolution with the active participation of users. Therefore, it deals with the analysis of software projects that have been under development and maintenance for several years and which are usually formed by thousands of software artifacts, which are also associated to logs from communications, defect-tracking and software configuration management systems. Accordingly, evolutionary visual software analytics aims to assist software developers and software project managers by means of an integral approach that takes into account knowledge extraction techniques as well as visual representations that make use of interaction techniques and linked views. Consequently, this paper discusses the implementation of an architecture based on the evolutionary visual software analytics process and how it supports knowledge discovery during software maintenance tasks.