Service B: The Aggregator

Service B is a resilient and scalable Java-based microservice responsible for consuming calculation results from Apache Kafka and maintaining a persistent, running total. It's designed with a clear separation of concerns for read and write operations using a Leader/Replica database pattern.

🚀 Core Functionality

• Kafka Consumer: Listens to the user-events topic for incoming messages generated by Service A's outbox pattern.
• Data Aggregation: Parses each message, extracts the calculated sum, and adds it to a running total.
• Persistent State: Stores the running total in a PostgreSQL database, ensuring data durability.
• Leader/Replica Pattern:
  • Write Operations: All updates to the running total are directed to a leader database (outbox_write), ensuring data consistency.
  • Read Operations: The API for retrieving the current total reads from a replica database (outbox_read), separating read and write workloads for better performance and availability.
• REST API: Exposes a GET /api/summation endpoint to allow clients to fetch the current aggregated sum.
• Observability: Integrated with Prometheus to export key metrics for monitoring service health and performance.

🛠️ Technology Stack

Category	Technology
Framework	Spring Boot 3.5
Language	Java 21
Messaging	Apache Kafka Consumer
Database	PostgreSQL (with Leader/Replica pattern)
API	REST (Spring Web)
Metrics	Micrometer, Prometheus

⚙️ How It Works

1. Message Consumption: The KafkaConsumerService listens for messages on the user-events topic.
2. Summation Update: Upon receiving a message, it calls the SummationService to coordinate the update.
3. Database Write: The DatabaseService updates the running total in the outbox table of the leader database.
4. Database Replication: The change is asynchronously replicated to the replica database. (Note: The current implementation simulates this by writing to the replica after the leader write is complete).
5. API Read: When a user hits the GET /api/summation endpoint, the SummationController fetches the value from the replica database, providing a fast, non-blocking read.

⚡ Load Testing

The service includes a K6 script to test the performance of its REST API.

• Test Script: k6_http.js.
• To run the test:
  1. Ensure the full application stack is running (make start in the Docker-Compose directory).
  2. Install K6.
  3. Navigate to the Service-B directory and execute the script:

  k6 run k6_http.js