Event-Driven & Streaming — Roadmap
Pub/sub patterns, CDC with Debezium, event sourcing, and schema evolution that doesn't break your consumers.
Real-World Analogy
A newspaper vs a phone call: a phone call (synchronous request/response) demands immediate attention from both parties. A newspaper (event stream) publishes facts once; readers consume them on their own schedule, and new readers can go back and read old editions. Event-driven systems give you the newspaper model — loose coupling, independent scaling, and a permanent record of what happened.
What you will learn
Events are how loosely coupled systems communicate. This track covers the vocabulary (events vs commands vs queries), the patterns for delivering events reliably (pub/sub, fan-out, consumer groups), how to capture database changes without touching application code (CDC with Debezium), the event sourcing pattern for audit-first systems, and how to evolve schemas over time without breaking deployed consumers.
Chapters in this track
- Events vs Commands vs Queries — three message types with different semantics, CQRS, naming conventions
- Pub/Sub Patterns — topics, consumer groups, fan-out, delivery guarantees, Kafka vs SNS/SQS
- Change Data Capture — Debezium, logical replication, the outbox pattern, monitoring slot lag
- Event Sourcing — append-only event store, aggregates, projections, snapshots, when to use it
- Schema Evolution — backward compatibility, Schema Registry, Avro, consumer-driven contracts