Title

AWS re:Invent 2022 - Smart serverless architecture for smart manufacturing (PRT301)

Summary

• Markus Grossmann from XL2 (a joint venture of Audi and Capgemini) presented on smart serverless architecture for smart manufacturing.
• The focus was on the layer above the shop floor, specifically systems and applications that control or receive data from the shop floor.
• A real use case of a parameter service developed for Audi was discussed, which transmits car-specific settings to production line stations, replacing a cumbersome USB drive process.
• Architectural challenges in manufacturing systems include the need for 24/7 availability, resilience, scalability, and security.
• Serverless architecture is a good fit for these challenges due to built-in high availability, scalability, event-driven workloads, integration services, and security.
• Principles and patterns for additional challenges in smart manufacturing were presented, including reusability, adaptability, and self-optimization.
• Component-oriented design and homogeneity are key for reusability and maintainability.
• Service stores are recommended for sharing reusable services within a company.
• Elastic data models are proposed for high adaptability, especially for UI changes during runtime without database schema changes.
• Event-driven architecture is essential for serverless applications and enables self-optimizing systems through closed feedback loops.
• Development principles for serverless projects include the importance of end-to-end tests over unit tests and treating infrastructure as code development with the same rigor as source code development.

Insights

• The session highlighted the importance of reusability and standardization in serverless architecture to avoid complexity and maintainability issues.
• The concept of a service store within an organization can facilitate the sharing and reuse of components, which can accelerate development and reduce costs.
• Elastic data models can provide the flexibility needed to adapt to changes in manufacturing processes without significant downtime or development effort.
• The emphasis on event-driven architecture and closed feedback loops suggests a move towards more autonomous and self-optimizing systems in manufacturing, which can improve efficiency and reduce human intervention.
• The recommendation to prioritize end-to-end testing in serverless projects reflects the complexity and interconnectedness of serverless components, where configuration issues can cause failures not caught by unit tests.
• Treating infrastructure as code with the same level of discipline as application code is a best practice that can lead to more reliable and maintainable serverless architectures.
• The session provided practical insights and examples from real-world applications, demonstrating the viability of serverless architecture in complex, high-stakes environments like automotive manufacturing.