Ten Best Practices for Event Streaming Success
summary
Ten concrete practices for event streaming success with examples: prioritize clear SLAs for ordering, delivery and latency; split domains and events instead of dumping into warehouses; avoid locking into one vendor by mixing Kafka Streams, Redpanda or Materialize; break transforms into small stages and show a real multi-topic aggregation pipeline that required RocksDB state stores and Kubernetes CSI to restore pods; prefer dumb pipes, smart endpoints and plan for replay complexity, stateful workloads, and who owns composite entities.
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:42Business Runs 24x7, so does Business Events
- 1:47Do we need Streaming if we have Storage?
- 2:33Terminology
- 3:23Event Streaming Approaches
- 3:57Setting Expectations
- 5:21You are not trying to build Legacy 2.0
- 6:18Be Aware of Business Gerunds
- 7:42Do not overinvest in a Single Platform
- 10:40Developer Upskilling
- 14:06Inefficient Topology can have lasting Consequences
- 19:09Handling Stateful Workloads
- 21:09Frequent Changing Needs Can Pose Challenges
- 23:01Centralized Ownership Can Be An Existential Threat
related talks
47mTop 10 Event-Driven Architecture Pitfalls
30mEvent Streaming and the Future of Applications
47mTop 10 Event-Driven Architecture Pitfalls
47mTackling Cloud Complexity - Recipes From Real World Projects
44mThe Art and Benefits of Computing Eventfully
51mComplex Event Flows in Distributed Systems
42mApache Flink Worst Practices
26mHow to Pick Your Next Streaming Architecture