# Comparing 2D and Augmented Reality Visualizations for Microservice System Understandability: A Controlled Experiment

**Authors:** Amr S. Abdelfattah, Tomas Cerny, Davide Taibi, Sira Vegas  
**arXiv:** [2303.02268](https://arxiv.org/abs/2303.02268)  
**Date:** 2023-03-03  
**Summary page:** https://www.summarizepaper.com/en/paper/2303.02268v1/

## Key points

- The study explores the use of augmented reality (AR) as a visualization tool for microservices in service-oriented design.
- The experiment followed a 2x2 crossover design where each participant group used both visualization tools - 2D and AR.
- The study aimed to prevent one tool from exerting more influence than the other by using abstract names for the two tools and randomly ordering the order of tool execution.
- Results show that the 3D visualization enables novices to perform as well as experts in detecting service dependencies, especially in large systems while no differences are reported for identifying service cardinality or bottlenecks.
- Due to compatibility issues with six participants' mobile devices not meeting minimum OS requirements, they struggled to use the AR tool resulting in a lack of AR tool results for four small and two large participants; therefore participant assignments were slightly adjusted.
- The study recommends industry and researchers further investigate AR for microservice architectural analysis, especially to ease the onboarding of new developers in microservice projects.

## Summary

This paper presents a study that explores the use of augmented reality (AR) as a visualization tool for microservices in service-oriented design. Microservice-based systems can be complex to understand, especially as they grow in size, and abstracted views can help practitioners gain system understanding from a certain perspective. Recent advancements in interactive data visualization have led to questions about whether established software engineering models remain the most suited approach for visualizing system design in microservices. The study proposes presenting a 3D visualization for microservices in AR and analyzes whether such an approach brings any benefits to practitioners when dealing with selected architectural questions related to system design quality. The experiment followed a 2x2 crossover design where each participant group used both visualization tools - 2D and AR. The participants were balanced based on system size and experience level, with ten small and ten large systems, and novice and experienced participants assigned to each system. The study aimed to prevent one tool from exerting more influence than the other by using abstract names for the two tools. Additionally, the order of tool execution was randomly ordered so that some participants completed the study in reverse order from others. The experiment duration was one hour, which included ten minutes for training, thirty minutes for executing the experiment and answering tasks related to identifying service dependency, service cardinality, and bottlenecks, and ten minutes for filling out feedback forms. The feedback step consisted of usage-related criteria questions using an ordinal 5-point Likert scale (1 is worst; 5 is best) and open-opinion questions that were analyzed to extract common thoughts and recommendations. Results show that the 3D visualization enables novices to perform as well as experts in detecting service dependencies, especially in large systems while no differences are reported for identifying service cardinality or bottlenecks. Due to compatibility issues with six participants' mobile devices not meeting minimum OS requirements (minimum Android 8 or iOS 11), they struggled to use the AR tool resulting in a lack of AR tool results for four small and two large participants; therefore participant assignments were slightly adjusted. In conclusion, the study recommends industry and researchers further investigate AR for microservice architectural analysis, especially to ease the onboarding of new developers in microservice projects. The assessment of the procedure indicated that all participants were able to carry out the 2D experiment successfully; nonetheless six participants could not install the AR tool due to compatibility issues with their mobile devices.