WebSockets ในสถาปัตยกรรมไมโครเซอร์วิส

Welcome to a deep dive into using WebSockets within a microservices architecture! Modern applications often use microservices for scalability and flexibility.

WebSockets are perfect for real-time features like chat, live updates, and notifications. But how do they fit into a distributed system?

Microservices are typically designed to be stateless. This means any instance of a service can handle any request, and no client connection state is stored directly on the service instance.

However, WebSockets are stateful. They maintain a persistent, long-lived connection between a client and a specific server instance. This creates a challenge in microservices.

Imagine you have multiple instances of your 'Chat Service'. If a client connects to Instance A, and later another backend service wants to send a message to that client, how does it know to reach Instance A?

• Load Balancing: Traditional load balancers might send subsequent HTTP requests to different service instances.
• Scaling: If Instance A goes down or scales, the connection is lost.
• Inter-service Communication: Other microservices can't directly talk to a specific client connection on another service.

To solve this, we introduce a dedicated WebSocket Gateway. This gateway is a microservice itself, specialized in handling all incoming WebSocket connections.

It acts as the single entry point for all real-time client traffic, managing the state of each active WebSocket connection.

The WebSocket Gateway takes on several crucial tasks:

• Connection Management: Handles WebSocket handshake and maintains all active client connections.
• Authentication: Often integrates with security to verify client identity.
• Message Routing: Forwards messages between clients and backend services.
• Presence: Can track which users are online.

Even with a gateway, how do backend services communicate with the gateway, especially if there are multiple gateway instances?

This is where an external Message Broker comes in. Brokers like RabbitMQ or Apache Kafka act as a central hub for all inter-service communication related to WebSockets.

Here's a simplified view of the architecture:

• Clients connect to the WebSocket Gateway.
• The Gateway and all Backend Microservices (e.g., Chat Service, Notification Service) communicate via a Message Broker.
• This decouples client connections from backend logic, allowing services to scale independently.

When a client sends a message:

1. Client sends message to the WebSocket Gateway.
2. Gateway receives it and publishes it to a specific topic/queue on the Message Broker (e.g., client.messages).
3. A relevant Backend Microservice (e.g., Chat Service) subscribes to this topic, receives the message, and processes it.

When a backend service wants to send a message to a client:

1. The Backend Microservice publishes the message to a topic/queue on the Message Broker (e.g., user.123.updates or a general server.broadcast).
2. The WebSocket Gateway subscribes to relevant topics and receives the message.
3. The Gateway identifies the target client(s) and forwards the message over the active WebSocket connection.

Messages exchanged between the Gateway and backend services via the broker need a clear structure. Here's a simple Java class representing such a message:

In a microservice architecture using WebSockets, which component is primarily responsible for managing individual client WebSocket connections?

We've explored how to integrate WebSockets into a microservices environment. Key takeaways:

• WebSockets' stateful nature conflicts with stateless microservices.
• A dedicated WebSocket Gateway manages client connections.
• A Message Broker (like RabbitMQ or Kafka) facilitates communication between the gateway and backend services.
• This architecture ensures scalability, decoupling, and robustness for real-time features in distributed systems.