Chat Application Architecture | AWS | API Gateway | WebSockets | Microservices | System Design

In this blog, we’ll dive into one of the ways to build a scalable chat application using AWS Cloud. No fluff—let’s get straight to it!

Functional Requirement

User should be able to chat using we or mobile app.

⚡🖥️Tech Stack

🌐 Front End
Website: 📄 HTML5 / 🟨 JavaScript (SPA) / ✨ Nextjs
Mobile App: 📱 Appropriate Tools

🤝 Interactions
All Client to Server interactions over 🔌 WebSockets
Request/response payload: 📦 JSON

⚙️ Backend
Backend will be 🚫 Serverless as far as possible
Deployed on ☁️ Cloud
Scale on demand: 📈

Logical Architecture

AWS Architecture

🧩 Components and Their Roles

1. Clients (Client 1, Client 2, ... n Clients)🖥️

   • These are WebSocket clients that connect to the API Gateway WebSocket API.
   • They send and receive messages through WebSockets.

2. API Gateway (WebSocket API)🚪

   • API Gateway manages WebSocket connections.
   • It provides endpoints for WebSocket connection, disconnection, and message handling.
   • It routes events to the appropriate AWS Lambda functions.

3. Lambda Functions🟧

   • Connect Lambda Function
     • Executes when a client establishes a WebSocket connection.
     • Stores the client’s connection ID in DynamoDB for message routing.
   • Disconnect Lambda Function
     • Triggers when a client disconnects.
     • Removes the connection ID from DynamoDB to prevent unnecessary message delivery.
   • onMessage Lambda Function
     • Handles incoming messages from WebSocket clients.
     • Retrieves connection IDs from DynamoDB to forward messages to the correct recipient.

4. DynamoDB🔷

   • Stores active WebSocket connections (connection IDs).
   • Used for tracking which clients are online and enables message delivery to the correct recipients.

How This 🏗️ Architecture Works

1. Client 1 and Client 2 connect to the WebSocket API via API Gateway.
2. API Gateway triggers the Connect Lambda function, which stores the connection ID in DynamoDB.
3. When a client sends a message, API Gateway invokes the onMessage Lambda function.
4. The onMessage Lambda function retrieves connection IDs from DynamoDB and forwards messages to the appropriate recipients using API Gateway.
5. When a client disconnects, API Gateway calls the Disconnect Lambda function, which removes the connection ID from DynamoDB.

How Far Can It Scale 📈?

• API Gateway supports up to millions of concurrent WebSocket connections.
• Lambda can handle thousands of requests per second (subject to AWS account limits).
• DynamoDB scales on-demand and can handle millions of transactions per second with proper indexing and partitioning.

🔚 Conclusion
This architecture is fully serverless, scalable, and cost-efficient for handling real-time WebSocket connections. It works well for medium to large-scale applications and can handle millions of users with minimal infrastructure management. 🚀