WebSockets & Real-Time Systems with Spring · レッスン

分散WebSocketアーキテクチャ

分散マイクロサービス環境でWebSocketアプリケーションを設計・実装します。

「分散WebSocketアーキテクチャ」はCoddyKit上の無料WebSockets & Real-Time Systems with Springレッスンです。これはレッスン3/4です。下記で完全なレッスンを無料で読むことができます。その後、ブラウザ内の組み込みコードエディタと24時間対応のAIチューターでハンズオン演習できます。これはWebSockets & Real-Time Systems with Spring学習パスの一部であり、ウェブとCoddyKitアプリ全体で進捗が同期されます。 WebSockets & Real-Time Systems with Springコースには全4レッスンが含まれています。

このレッスンの一部はまだ翻訳されておらず、英語で表示されています。

Why Distribute WebSockets?

As your application grows, a single WebSocket server might not be enough to handle all user connections and message traffic.

Distributed WebSocket architectures allow you to scale your real-time applications by running multiple server instances. This helps with:

Load balancing: Spreading connections across servers.
High availability: No single point of failure.
Microservices: Integrating real-time features into a distributed system.

The Stateful Challenge

A core challenge with WebSockets is their stateful nature. Each client maintains a persistent connection with a specific server instance.

If you have multiple server instances (Server A, Server B), and a client connected to Server A sends a message meant for a client connected to Server B, how does Server A know where to send it?

This problem requires a way for server instances to communicate with each other.

Load Balancers & Sticky Sessions

To distribute incoming WebSocket connections, you'll use a load balancer (e.g., NGINX, HAProxy).

It directs new connection requests to one of your available WebSocket server instances.
For WebSockets, it's common to use sticky sessions (also called session affinity). This ensures that once a client connects to a specific server instance, all subsequent messages for that WebSocket connection are routed to the same instance.

This keeps the stateful connection intact between the client and its assigned server.

External Brokers Connect Instances

While sticky sessions handle client-to-server routing, we still need servers to talk to each other. This is where external message brokers become crucial.

Recall from previous lessons: brokers like RabbitMQ or Kafka act as a central communication hub. In a distributed setup:

Server instances publish messages to the broker.
Other server instances subscribe to topics on the broker and consume messages.

This allows messages to be efficiently broadcast or routed between any server instance.

Broadcasting Across the Cluster

Imagine you have a chat room. When a user sends a message, it needs to reach everyone in that room, even if they're connected to different server instances.

Here's how it works:

A client sends a message to its connected server instance (e.g., Server A).
Server A publishes this message to a specific topic on the external message broker.
All other server instances (Server B, Server C, etc.) subscribe to that same topic on the broker.
When they receive the message from the broker, they forward it to their respective connected clients who are in that chat room.

Example: Distributed Broadcast

This simple example simulates a server instance publishing a message to a topic. In a real Spring application, you'd use a SimpMessagingTemplate to send to the external broker.

Try running this example:

public class MessagePublisher {
    public static void main(String[] args) {
        String message = "User joined room 'general'!";
        String destination = "/topic/chat/general";

        System.out.println("--- Distributed Message System ---");
        System.out.println("Server instance publishing message:");
        System.out.println("Destination: " + destination);
        System.out.println("Content: \"" + message + "\"");
        System.out.println("\n(This message would be sent to an external broker,");
        System.out.println("then routed to all connected clients subscribing");
        System.out.println("to " + destination + " across all server instances.)");
    }
}

Targeting Users in a Cluster

What if you want to send a private message to a specific user, regardless of which server instance they're connected to?

With STOMP, you can use user-specific destinations (e.g., /user/{username}/queue/private-messages). When a server publishes to such a destination:

The external broker identifies which server instance the target user is connected to.
The broker then routes the message directly to that specific instance.
That instance then delivers the message to the user's private queue.

This abstracts away the complexity of knowing the user's exact server instance.

Service Discovery in Action

In a truly dynamic, distributed environment (like microservices), server instances come and go. How do they find each other or register their presence?

Service discovery tools (e.g., Netflix Eureka, Consul) help:

Each WebSocket server instance registers itself with a discovery service upon startup.
Other services can query the discovery service to find available WebSocket instances.

While not directly handling WebSocket traffic, service discovery is vital for managing the dynamic nature of distributed server clusters.

Scaling Best Practices

To build robust distributed WebSocket applications:

Horizontal Scaling: Add more WebSocket server instances as traffic grows.
Externalize State: Avoid storing user or session-specific data directly on the WebSocket server instances. Use external databases, caches (like Redis), or the message broker for shared state.
Stateless Logic: Design your application logic to be as stateless as possible, making it easier to scale.
Monitoring: Keep a close eye on connection counts, message rates, and server health across all instances.

Distributed Architecture Quiz

In a distributed WebSocket architecture, what is the primary role of an external message broker like RabbitMQ or Kafka?

Distributed WebSockets Recap

Great job! You've learned about designing and implementing distributed WebSocket applications:

Why distribute: Scaling, high availability, microservices.
Challenges: Stateful connections, inter-server communication.
Solutions: Load balancers with sticky sessions, external message brokers for inter-instance messaging.
Patterns: Broadcasting to all clients, targeting specific users via brokers.
Support: Service discovery for managing dynamic instances.

These principles are key to building robust and scalable real-time systems!

よくある質問

「分散WebSocketアーキテクチャ」レッスンは無料ですか？

はい。「分散WebSocketアーキテクチャ」の完全なテキストはこのウェブで無料で読めます。インタラクティブに演習し（組み込みコードエディタと24時間対応のAIチューター）、WebSockets & Real-Time Systems with Springコースの残りをアンロックするには、CoddyKit PROにアップグレードしてください。 WebSockets & Real-Time Systems with Springコースには全4レッスンが含まれています。

「分散WebSocketアーキテクチャ」で何を学びますか？

分散マイクロサービス環境でWebSocketアプリケーションを設計・実装します。ブラウザで直接実行するハンズオンコードでWebSockets & Real-Time Systems with Springを演習し、24時間対応のAIチューターがレッスンを進める中での質問に答えます。

WebSockets & Real-Time Systems with Springを始めるのに経験は必要ですか？

事前経験は必要ありません。CoddyKitのWebSockets & Real-Time Systems with Springは初級者から上級者向けに構成されているため、ここから始めるか最初から始めて、自分のペースで進むことができます。これはレッスン3/4です。

「分散WebSocketアーキテクチャ」レッスンにはどのくらい時間がかかりますか？

ほとんどのCoddyKitレッスンは約5～10分かかります。各レッスンはコンパクトでインタラクティブなので、着実に進歩し、ウェブとアプリ全体で正確に前回の場所から再開できます。

このWebSockets & Real-Time Systems with Springレッスンでコードを書いて実行できますか？

はい。すべてのWebSockets & Real-Time Systems with Springレッスンに組み込みコードエディタが含まれているため、ブラウザでリアルコードを書いて実行し、即座のAIフィードバックを取得できます。ローカル設定は不要です。

このコースのすべてのレッスン

← WebSockets & Real-Time Systems with Springに戻る