An Architecture for Self-organizing

Continuous Delivery Pipelines

Attracted by the competitive advantages of being able to release software quickly and reliable, many organizations have adopted Continuous Delivery (CD) practices in recent years.

However, both academia and industry have reported on several adoption challenges. One major challenge are the infrastructure and tools supporting Continuous Delivery with existing delivery systems being inflexible and require lots of technical and process-related knowledge.

This thesis introduces a microservice-based architecture for flexible and maintainable delivery systems which provides means to reduce the knowledge required through analyzing, complementing and optimizing the delivery model. While at the same time providing extensive validation support.

A case study in an industrial context showed promising results and confirmed the practicability of our approach. The prototype’s delivery process adoptions thereby provided up to 3,31 times faster feedback times as compared with the manually modeled process. Overall, the architecture provides a flexible foundation for further research in various areas, ranging from modeling tools with auto-completion and recommendation support over smell and anti-pattern detection mechanisms to the development of a holistic delivery ecosystem.

Supervisor

Resources

• Thesis PDF
• Slides Endpresentation