Kafka with Spring Boot: Make ShopStream Event-Driven

Why This Course?

ShopStream's order-service calls fulfillment over HTTP: when fulfillment is slow, checkout is slow; when it is down, orders fail. Most Kafka tutorials stop at "send a message". This one breaks things on purpose — a stopped consumer, a wrong partition key, a slow consumer that triggers a rebalance storm, a broker killed mid-write, a crash between a database commit and a publish — and measures what each failure costs in lost, duplicated or out-of-order orders. Then you fix each one with the setting or pattern production teams use.

What You'll Build

Two Spring Boot 3.5 services — order-service and fulfillment-service — talking through Apache Kafka 3.9 in KRaft mode: keyed order events on a 6-partition topic, a consumer group of three, a 3-broker cluster with acks=all and an idempotent producer, a transactional outbox backed by a database, a dead-letter topic and a lag alert. Every lesson ends with a demo that prints real numbers and a verify script that gives you a completion code.

Who Should Take This Course?

• Backend engineers who use Spring Boot and want to use Kafka without surprises.

• Engineers preparing for system design interviews who need to explain ordering, delivery guarantees and the outbox pattern with evidence.

• Anyone who finished "Scale ShopStream" (step 1 of the free path) and wants the event-driven version.

What Makes This Course Different?

No Docker, no cloud account: each lesson's start.sh downloads Kafka once and runs it locally or in the browser sandbox. Each lesson has a build, a deliberate break and a measurement — for example, killing the leader broker during 20,000 orders: acks=1 lost 65 acknowledged orders, acks=all without idempotence delivered 449 twice, and the safe producer lost and duplicated none.