Choreography and Orchestration: React to Kafka Events in Real-Time with Kafka Connect and Business-as-Code

While event-driven choreography and imperative workflow orchestration are often framed as competing architectures, in practice most production systems need both. Teams who lean too far in either direction run into trouble: chaining microservices together with queues to implement a single business process produces a sprawling, hard-to-debug mess, while building behemoth workflows that span unrelated business domains creates a fragile, distributed monolith.

This talk presents practical, battle-tested guidelines for balancing orchestration and choreography in real production Kafka environments:

1. Using Business-as-Code to orchestrate multi-step processes within a single domain, keeping logic cohesive, debuggable, and easy to evolve.
2. Publishing events to Kafka once workflows reach important milestones or completion, so other domains can react without tight coupling.
3. Using Kafka Streams or Flink to enrich, join, and aggregate events from multiple sources—including running workflows—to build real-time views and trigger downstream processes.

Finally, a live demo with Kafka Connect, Kafka Streams, and the LittleHorse Server will leave the audience with a concrete, end-to-end example of how a balanced approach to choreography and orchestration helps you tame complexity and get the most out of your event-driven systems.