The InfoQ Podcast Event Driven Architectures of Scale
52 snips
Nov 19, 2021 Gwen Shapira, a Principal Engineer at Confluent and a Kafka expert, and Ian Thomas, Senior Principal Engineer at Flutter International, share their insights on the evolution and complexities of event-driven architectures (EDA). They discuss the transition from monolithic systems to EDA, emphasizing the importance of trust and domain models. The conversation highlights Kafka topic design, the significance of observability, and best practices for managing upgrades. Their experiences in the betting and gaming industry reveal the unique challenges and advantages of implementing EDA at scale.
AI Snips
Chapters
Transcript
Episode notes
Partition Choices Have Long-Term Cost
- Early partition choices can become irreversible and hinder scaling later.
- Ian Thomas highlights compacted topics and partition counts as long-term constraints to design for.
Event Versioning Is Hard
- Event versioning is a deep, difficult topic that deserves focused attention.
- Gwen Shapira notes an entire book exists because evolving events safely is non-trivial.
Invest In Upgrade-Test Frameworks Early
- Build fast, reliable test frameworks before you deploy so upgrades don't break systems.
- Gwen Shapira urges daily smoke checks and tests that run in reasonable time to build confidence in releases.