Event Driven Architecture (EDA) in Microservices

1. What is Event Driven Architecture?

Event Driven Architecture হলো এমন একটা pattern যেখানে services একে অপরের সাথে events এর মাধ্যমে communicate করে।

• Event = কোন কিছু ঘটেছে এরকম একটা notification (e.g., "User Registered", "Order Placed", "Payment Completed")
• Services events produce করে এবং অন্য services সেই events consume করে
• Asynchronous communication - services একে অপরের জন্য wait করে না

2. Event vs Traditional API Call

Traditional Synchronous Call:

Order Service → (API Call) → Payment Service → Response
               ↓ (Waiting...)
            (Blocked until response)

Event Driven Approach:

Order Service → (Event: "Order Created") → Message Queue
                                               ↓
                                        Payment Service (Listens)
                                        Notification Service (Listens)
                                        Inventory Service (Listens)

3. Key Components of EDA

A. Event Producer

• যে service event তৈরি করে এবং publish করে
• Example: Order Service যখন order create করে তখন "OrderCreated" event produce করে

B. Event Consumer

• যে service event listen করে এবং তার উপর ভিত্তি করে action নেয়
• Example: Payment Service "OrderCreated" event শুনে payment process করে

C. Event Broker/Message Queue

• Events গুলো store করে এবং deliver করে
• Popular tools: Apache Kafka, RabbitMQ, AWS SQS, Redis Pub/Sub

D. Event Store

• সব events history রাখার জন্য (Event Sourcing এর জন্য)

4. Types of Events

A. Domain Events

• Business domain এর important events
• Example: "UserRegistered", "OrderPlaced", "PaymentCompleted"

B. Integration Events

• Services এর মধ্যে communication এর জন্য
• Example: "SendNotification", "UpdateInventory"

C. System Events

• Technical/Infrastructure events
• Example: "ServiceStarted", "DatabaseConnected"

5. Event Flow Example (E-commerce)

1. User places order
   ↓
2. Order Service → Publishes "OrderCreated" event
   ↓
3. Event gets stored in Message Queue (Kafka)
   ↓
4. Multiple services consume the event:
   - Payment Service → Process payment
   - Inventory Service → Update stock  
   - Notification Service → Send confirmation email
   - Analytics Service → Track order metrics

Real Event Structure:

{
  "eventId": "evt_12345",
  "eventType": "OrderCreated", 
  "timestamp": "2023-10-20T10:30:00Z",
  "source": "OrderService",
  "data": {
    "orderId": "order_789",
    "userId": "user_456", 
    "items": [...],
    "totalAmount": 1500
  }
}

6. Advantages of Event Driven Architecture

✅ Loose Coupling → Services একে অপরের সাথে directly connected না, event এর মাধ্যমে communicate করে

✅ Asynchronous Processing → Services parallel এ কাজ করতে পারে, কেউ কারো জন্য wait করে না

✅ Scalability → Event consumer গুলো independently scale করা যায়

✅ Fault Tolerance → একটা service down থাকলেও event queue তে stored থাকবে, later process হবে

✅ Real-time Processing → Events immediately broadcast হয়, real-time updates পাওয়া যায়

✅ Extensibility → নতুন service add করতে চাইলে শুধু event listen করলেই হবে, existing code change করতে হবে না

7. Disadvantages of Event Driven Architecture

❌ Complex Debugging → Event flow trace করা কঠিন, কোন event কোথা থেকে এসেছে বুঝা hard

❌ Eventual Consistency → Data immediately consistent না, eventually consistent হয়

❌ Message Queue Dependency → Message broker down হলে communication stop

❌ Duplicate Events → Same event multiple times process হতে পারে (idempotency handle করতে হয়)

❌ Order of Events → Events এর order maintain করা challenging

❌ Testing Complexity → Integration testing complicated

8. Event Sourcing Pattern

Event Sourcing হলো এমন একটা pattern যেখানে:

• Database এ current state store না করে sequence of events store করা হয়
• Current state events replay করে বের করা হয়

Example:

Traditional Database:
User Table: {id: 1, name: "John", email: "john@example.com", status: "active"}

Event Sourcing:
Event Store: 
- UserCreated {id: 1, name: "John", email: "john@example.com"}
- UserEmailUpdated {id: 1, newEmail: "john.doe@example.com"}  
- UserActivated {id: 1}

Benefits: Complete audit trail, time travel queries, easy rollback

9. CQRS (Command Query Responsibility Segregation)

Event Driven Architecture এর সাথে CQRS pattern often ব্যবহার হয়:

• Command Side → Write operations, events produce করে
• Query Side → Read operations, events consume করে read models update করে

Commands → Write DB → Events → Read Models → Queries

10. Message Queue Technologies

A. Apache Kafka

• High throughput, distributed streaming platform
• Event sourcing এর জন্য perfect
• Real-time data streaming

B. RabbitMQ

• Traditional message broker
• Reliable delivery, routing flexibility
• AMQP protocol

C. AWS SQS/SNS

• Cloud managed service
• SQS = Queue, SNS = Pub/Sub
• Serverless integration

D. Redis Pub/Sub

• In-memory, very fast
• Simple pub/sub messaging
• Good for real-time features

11. Event Patterns

A. Pub/Sub (Publish-Subscribe)

Publisher → Topic → Multiple Subscribers

B. Point-to-Point

Sender → Queue → Single Receiver

C. Request-Reply

Service A → Request Event → Service B → Reply Event → Service A

12. Best Practices

A. Event Design

• Events should be immutable
• Include all necessary data (avoid chatty calls)
• Use clear naming conventions
• Version your events

B. Idempotency

• Handle duplicate events gracefully
• Use unique event IDs
• Implement idempotent consumers

C. Error Handling

• Dead letter queues for failed events
• Retry mechanisms with exponential backoff
• Circuit breaker pattern

D. Monitoring

• Track event flow and latency
• Monitor queue sizes
• Alert on failed events

13. Real-world Examples

Netflix

• User action events → Recommendation engine updates
• Video streaming events → Analytics and monitoring

Uber

• Ride request events → Driver matching, pricing, ETA calculation
• Trip completion events → Payment, rating, analytics

Spotify

• Play events → Recommendation algorithms
• User interaction events → Playlist suggestions

14. Implementation Example (Node.js + Kafka)

Event Producer:

// Order Service
const kafka = require('kafkajs');
 
async function createOrder(orderData) {
  // Save order to database
  const order = await saveOrder(orderData);
  
  // Publish event
  await publishEvent('OrderCreated', {
    orderId: order.id,
    userId: order.userId,
    items: order.items,
    total: order.total
  });
}

Event Consumer:

// Payment Service
const consumer = kafka.consumer({ groupId: 'payment-service' });
 
await consumer.subscribe({ topic: 'OrderCreated' });
 
await consumer.run({
  eachMessage: async ({ message }) => {
    const event = JSON.parse(message.value.toString());
    
    if (event.eventType === 'OrderCreated') {
      await processPayment(event.data);
    }
  }
});

15. When to Use Event Driven Architecture

✅ Use When:

• Microservices architecture
• Real-time data processing needed
• High scalability requirements
• Complex workflows with multiple services
• Need for audit trails
• Decoupled system design

❌ Avoid When:

• Simple CRUD applications
• Strong consistency requirements
• Small team/simple architecture
• Immediate response needed (synchronous)

16. EDA vs API-based Communication

17. Migration Strategy

Step 1: Identify Events

• Find domain events in your system
• Start with high-value, low-risk events

Step 2: Choose Message Broker

• Kafka for high throughput
• RabbitMQ for reliability
• Cloud services for simplicity

Step 3: Implement Gradually

• Start with non-critical flows
• Add monitoring and alerting
• Gradually migrate critical paths

Step 4: Handle Dual Writes

• Use outbox pattern or CDC (Change Data Capture)
• Ensure data consistency during migration

18. Summary

Event Driven Architecture microservices এর জন্য অনেক powerful pattern:

Key Benefits:

• Services decoupled থাকে
• Asynchronous processing
• Better scalability এবং fault tolerance
• Real-time capabilities

Key Challenges:

• Debugging complexity
• Eventual consistency
• Infrastructure overhead

Perfect for: Large scale, complex systems যেখানে real-time processing এবং loose coupling দরকার।