Immersive Visualization of CI/CD Pipelines

Abstract

As CI/CD pipelines become increasingly complex, developers struggle to understand workflow structures, track failures, and navigate execution histories. Traditional tools like GitHub Actions use textual logs and 2D dashboards that can make it challenging to visualize temporal relationships and concurrency patterns across multiple pipeline runs.

This dissertation explores how Virtual Reality can improve CI/CD workflow visualization. Using a factory metaphor, pipeline data is translated into immersive 3D environments: workflows become floors, with their corresponding workflow runs represented as blocks placed on each floor, jobs become sub-blocks within those runs, and time unfolds as spatial depth. Users can walk through the pipeline history and understand execution patterns through natural movement.

A working VR prototype was implemented using Unity that transforms GitHub Actions data into an interactive 3D environment. The system includes temporal mapping algorithms, parallel and sequential layout organization for concurrent and dependent runs and jobs, time compression for inactive periods, and spatial audio for failure detection. Users can explore workflow history, identify problems, and understand pipeline behavior through spatial navigation.

A user study compared the VR system against GitHub Actions’ web interface using tasks for temporal navigation, failure detection, and debugging. In VR, users completed tasks more quickly and achieved higher success rates compared to when using the GitHub Actions interface. The VR system received higher usability scores, with participants finding it easier to use and less complex.

This research provides practical guidelines for immersive CI/CD visualization. It demonstrates that spatial 3D layouts help with overview tasks and temporal navigation, while traditional interfaces work better for detailed analysis.