Kafka KRaft practice Spring Boot microservices

# Practical Kafka Implementation in Spring Boot Microservices

Mid-to-senior developers can quickly deploy a Kafka cluster using KRaft and integrate it with four Spring Boot services for an online store. The order-service accepts orders via REST API and publishes an order-placed event. The inventory-service reserves stock and generates inventory-reserved or inventory-rejected events. The analytics-service and notification-service process these events asynchronously.

Requirements: Java, Spring Boot, Docker, Docker Compose, Postman. Dependencies: Spring for Apache Kafka, Spring Web.

Docker Compose for Kafka Cluster

Deploying three Kafka 8.1.0 brokers with KRaft without Zookeeper. Each broker performs both broker and controller roles. Kafka-UI on port 8086 for monitoring topics, partitions, and consumer groups.

Create docker-compose.yaml:

services:
  kafka-0:
    image: confluentinc/cp-kafka:8.1.0
    container_name: kafka-0
    hostname: kafka-0
    ports:
      - "29092:29092"
    environment:
      CLUSTER_ID: ZETVJENWRjaECDUEExP_eg
      KAFKA_NODE_ID: 0
      KAFKA_PROCESS_ROLES: broker,controller
      KAFKA_LISTENERS: INTERNAL://:9091,EXTERNAL://:29092,CONTROLLER://:9093
      KAFKA_CONTROLLER_QUORUM_VOTERS: 0@kafka-0:9093,1@kafka-1:9093,2@kafka-2:9093
      KAFKA_ADVERTISED_LISTENERS: INTERNAL://kafka-0:9091,EXTERNAL://localhost:29092
      KAFKA_LISTENER_SECURITY_PROTOCOL_MAP: INTERNAL:PLAINTEXT,EXTERNAL:PLAINTEXT,CONTROLLER:PLAINTEXT
      KAFKA_CONTROLLER_LISTENER_NAMES: CONTROLLER
      KAFKA_INTER_BROKER_LISTENER_NAME: INTERNAL
  # Similarly for kafka-1 (port 29093), kafka-2 (port 29094)
  kafka-ui:
    image: provectuslabs/kafka-ui:v0.7.2
    container_name: kafka-ui
    ports:
      - "8086:8080"
    environment:
      KAFKA_CLUSTERS_0_NAME: local
      KAFKA_CLUSTERS_0_BOOTSTRAPSERVERS: kafka-0:9091,kafka-1:9091,kafka-2:9091
      DYNAMIC_CONFIG_ENABLED: true

Startup command: docker compose up -d. The cluster is ready for load with 3 partitions and x3 replication.

Spring Boot Services Configuration

Common Producer Settings

In application.properties for all services:

spring.kafka.bootstrap-servers=localhost:29092,localhost:29093,localhost:29094
spring.kafka.producer.key-serializer=org.apache.kafka.common.serialization.StringSerializer
spring.kafka.producer.value-serializer=org.springframework.kafka.support.serializer.JsonSerializer

Programmatic Topic Creation

KafkaConfig class in the config package:

@Configuration
public class KafkaConfig {
    @Bean
    public NewTopic orderPlacedTopic() {
        return TopicBuilder.name("order-placed")
            .partitions(3)
            .replicas(3)
            .config(TopicConfig.RETENTION_MS_CONFIG, "86400000")
            .config(TopicConfig.RETENTION_BYTES_CONFIG, "524288000")
            .build();
    }
}

Retention: 24 hours, 500MB per topic. Similarly for inventory-reserved, inventory-rejected in inventory-service.

Order-Service Implementation (Producer)

DTOs and Events

public record OrderRequest(String email, String productName, Integer quantity) {}

public record OrderPlacedEvent(String orderId, String email, String productName, Integer quantity) {}

Business Logic

@Service
public class OrderService {
    private final KafkaTemplate<String, OrderPlacedEvent> kafkaTemplate;

    public OrderService(KafkaTemplate<String, OrderPlacedEvent> kafkaTemplate) {
        this.kafkaTemplate = kafkaTemplate;
    }

    public String placeOrder(OrderRequest orderRequest) {
        String orderId = UUID.randomUUID().toString();
        OrderPlacedEvent event = new OrderPlacedEvent(orderId, orderRequest.email(), 
            orderRequest.productName(), orderRequest.quantity());
        kafkaTemplate.send("order-placed", event);
        return orderId;
    }
}

The REST controller returns 201 Created with the orderId instantly, with processing happening asynchronously.

Inventory-Service: Producer + Consumer

Listens to order-placed, checks stock levels, and publishes inventory-reserved or inventory-rejected. Requires @KafkaListener and a second KafkaTemplate.

Service roles:

• Pure Producer: order-service
• Pure Consumer: analytics-service, notification-service
• Producer+Consumer: inventory-service

Consumer Implementation (analytics, notification)

@KafkaListener(topics = "inventory-reserved")
public void handleInventoryReserved(InventoryReservedEvent event) {
    // Analytics logic
}

Spring Kafka automatically manages offsets, with consumer groups formed by service name.

Testing the Flow

• POST /orders with JSON { "email": "[email protected]", "productName": "laptop", "quantity": 1 }
• Monitor in Kafka-UI: order-placed → inventory-reserved
• Analytics and notification receive events in parallel

Key Points:

• KRaft simplifies deployment without Zookeeper
• 3 partitions + x3 replication = fault tolerance
• JsonSerializer for value, StringSerializer for key
• 24h/500MB retention suitable for development
• KafkaTemplate abstracts low-level API

— Editorial Team