🚀 Overview

This PR implements a transaction processing system using an event-driven architecture.

The solution introduces asynchronous communication between services using Kafka, ensuring decoupling, scalability, and resilience.

🧩 Architecture

• Transaction Service publishes events when a transaction is created
• Anti-Fraud Service consumes events and evaluates transactions
• Transaction status is updated asynchronously based on the validation result

⚙️ Key Features

Event-Driven Architecture

• Implemented using Kafka for asynchronous communication
• Decouples transaction processing from fraud validation

Idempotency

• HTTP layer: uses Idempotency-Key to prevent duplicate transactions
• Consumer layer: prevents duplicate event processing using Redis

Redis Integration

• Used for:

  • Idempotency key storage
  • Deduplication of Kafka events

• TTL applied to avoid unbounded growth

Transaction Lifecycle

• PENDING → APPROVED | REJECTED

🛠️ Tech Stack

• Node.js (NestJS)
• Kafka (event streaming)
• Redis (idempotency + deduplication)
• PostgreSQL (persistence)

🧪 How to Test

1. Start services:

docker-compose up -d

2. Run both services

3. Create a transaction:

POST /transactions
Header: Idempotency-Key: <unique-key>

4. Verify:

• Transaction is created with PENDING
• Anti-fraud processes event
• Status updates to APPROVED or REJECTED

📌 Notes

• System is eventually consistent
• Designed to handle service failures (Kafka retains events)
• Idempotency ensures safe retries