Mining Container Image Repositories for Software Configuration and Beyond
Citation: Tianyin Xu, Darko Marinov Mining Container Image Repositories for Software Configuration and Beyond.
Internet Archive Scholar (search for fulltext): Mining Container Image Repositories for Software Configuration and Beyond
Wikidata (metadata): Q64336637
Download: https://arxiv.org/pdf/1802.03558.pdf
Tagged:
Summary
Advocates studying container image repositories, which can reveal information about use of software that previously could only be gathered through expensive and biased interviews because there was no reason for configuration information to be shared before. Opportunities include:
- Creating a feedback loop for configuration design
- Modeling cross-component configuration dependencies
- Discovering software orchestration
- Improving combinatorial testing and tuning
- Using images as test beds for software engineering tools
Challenges include that container image repositories are large and images are created using heterogeneous conventions. Outlines potential methodologies including streaming, downloading, and studying layers.