Capturing and Visualizing Data Accesses in Microservices Architectures

Défense de mémoire de Maxime De Rycke

Over the last years, microservices architectures have gained popularity over monolithic systems, including in large companies such as Google, Netflix, or Spotify.
This architectural style is typically associated with qualities such as modularity, resilience, composability, interoperability, and heterogeneity. However, the distributed
and heterogeneous nature of these architectures can complicate the task of holistic understanding of the system for developers. This problem also occurs at the level
of different databases and their access. The literature suggests different techniques for reconstructing and understanding microservices architectures, including from a
data access perspective, using static or dynamic analysis, but less commonly a combination of both. This thesis aims to fill this gap by suggesting a new approach,
consisting of two contributions. First, it extends an existing static analysis technique with heuristics for data access. Its goal is to supplement it with additional information derived from dynamically collected traces. Then, it focuses on facilitating the understanding of data access by representing it through two interactive visualizations. Initially, in the form of a timeline. Subsequently, as an animated treemap
highlighting frequently used components of the architecture. Both contributions are evaluated in a case study involving Overleaf, on which a scenario spanning several
days is observed and finally commented upon.

Keywords : microservices, data access, reverse engineering, visualization

Contact : Isabelle Daelman - isabelle.daelman@unamur.be
Télecharger : vCal