🎉 follow my github →
Backend
Application Architecture
Serverless Architecture

Serverless Architecture

1. What is Serverless Architecture?

Serverless Architecture হলো এমন একটা design pattern যেখানে developer কে server manage করার চিন্তা করতে হয় না।

  • Server থাকে, কিন্তু পুরোপুরি cloud provider managed।
  • Developer শুধু code লিখে deploy করে।
  • Automatic scaling, patching, maintenance সব cloud provider করে।

2. Core Principles of Serverless

  1. Function-as-a-Service (FaaS) → কোডকে ছোট ছোট function হিসেবে deploy করা হয়।
  2. Event-Driven → Function trigger হয় HTTP request, DB event, message queue, cron job ইত্যাদি দিয়ে।
  3. No Server Management → Server, VM, OS সব cloud provider manage করে।
  4. Auto-Scaling → Traffic অনুযায়ী functions auto-scale করে।
  5. Pay-as-you-go → শুধুমাত্র code execution time ও resource ব্যবহার অনুযায়ী pay।

3. Structure of Serverless Applications

  • Client / UI → Browser, Mobile app
  • API Gateway / HTTP Trigger → Request route করে
  • Serverless Functions → AWS Lambda / Azure Functions / Google Cloud Functions
  • Databases / Storage → Cloud-managed DB (DynamoDB, Firebase, CosmosDB)
  • Message Queue / Event Bus → Async trigger (SQS, Kafka, EventBridge)

Example: Photo Upload App

  1. User uploads photo → triggers Lambda function
  2. Lambda function → Resizes photo → Stores in S3 / Cloud Storage
  3. Event → Another function → Sends email notification

4. How Serverless Works

ধরা যাক আপনি একটি HTTP API বানাচ্ছেন :

  1. User → HTTP request → API Gateway
  2. API Gateway → triggers Lambda function
  3. Function executes business logic → Accesses DB, returns response
  4. Function finish → Resources auto-released
  5. সব auto-scale হয় traffic অনুযায়ী

5. Advantages of Serverless

✅ No Server Management → Server patching, scaling handled by cloud ✅ Automatic Scaling → Zero to thousands of requests handled automatically ✅ Cost Efficiency → Pay only for actual execution time ✅ Fast Deployment → Function-level deployment, small units ✅ Focus on Code → Devs focus on business logic, not infra

6. Disadvantages

❌ Cold Start Latency → First request after idle may be slow ❌ Vendor Lock-in → Code tightly coupled with cloud provider ❌ Limited Execution Time → Functions usually have timeout limits ❌ Debugging & Testing → Local testing harder ❌ Complex for Large Apps → Too many functions = management overhead

7. When to Use Serverless

  • Lightweight APIs / microservices
  • Event-driven apps (file upload, notifications, IoT triggers)
  • Applications with spiky or unpredictable traffic
  • Rapid prototyping or MVP

8. When NOT to Use

  • Long-running processes ( > 15 min in AWS Lambda)
  • High-performance / low-latency apps
  • Heavy computation apps (ML training, video rendering)
  • Complex stateful apps (requires session management)

9. Real-Life Examples

  • Netflix → Serverless for notifications, image processing
  • Coca-Cola Freestyle → Serverless for IoT vending machines
  • Airbnb → Serverless functions for lightweight APIs
  • iRobot → Robot cloud functions for device events

10. Scaling in Serverless

  • Automatic & Infinite Scaling → Functions scale per request
  • No manual intervention needed
  • Pay per execution → Cost optimized

11. Deployment in Serverless

  • FaaS Platforms → AWS Lambda, Azure Functions, Google Cloud Functions
  • Triggers → API Gateway, Cron Jobs, DB Event, S3 Event
  • CI/CD → Serverless Framework, SAM, Terraform, GitHub Actions

12. Important Concepts

  • Event-Driven → Functions respond to events, not always-on server
  • Stateless Functions → No persistent local state
  • Managed Services → DB, Queue, Storage managed by cloud
  • Observability → Cloud provider tools (CloudWatch, Stackdriver)

13. Diagram (Textual Representation)

        User
         |
      API Gateway
         |
 -----------------------------
| Function 1 | Function 2     |
| (HTTP API) | (DB Trigger)   |
|------------|----------------|
       |          |
      DB        Storage / Queue
  • Functions triggered by events
  • Auto-scale & server handled by cloud

14. Developer Experience

  • Beginners → Easy to start, no infra setup needed
  • Intermediate → Need to learn event-driven design
  • Large Team → Can split logic into small functions, maintain easily

15. Summary

  • Serverless = No server management
  • Event-driven, auto-scaled, pay-as-you-go
  • Perfect for lightweight, spiky traffic apps
  • Not suitable for heavy compute or long-running tasks
Cloud-native ArchitectureMonorepo Architecture