Redis 发布/订阅机制

Welcome back! In the previous lesson, we learned what Redis Pub/Sub is. Now, let's dive into the mechanics of how it actually works inside Redis.

We'll explore the main components: channels, publishers, and subscribers. Understanding these roles is key to building real-time features.

At the heart of Redis Pub/Sub are channels. Think of a channel as a named topic or a broadcast radio station.

• Named Conduits: Channels are simply names (strings) that messages are sent to and received from.
• No Storage: Redis channels themselves do not store messages. They are purely for real-time message routing.
• Dynamic Creation: Channels are created implicitly when the first client publishes to or subscribes to them.

A publisher is any client that sends a message to a specific channel. Publishers don't need to know who (if anyone) is listening.

• Fire and Forget: Publishers send a message and immediately move on. They don't wait for acknowledgements.
• Decoupled: Publishers are completely decoupled from subscribers. They just need the channel name and the message content.

This decoupling is a major benefit for scalable, real-time applications.

A subscriber is a client that expresses interest in receiving messages from one or more specific channels.

• Active Listening: Subscribers actively listen to channels. When a message is published to a subscribed channel, Redis pushes it to the subscriber.
• Client-Side Logic: Subscribers usually have application logic to process the incoming messages.

Subscribers must be connected to Redis to receive messages. If a subscriber disconnects, it misses any messages published during its offline period.

Let's visualize the simple flow:

1. A publisher sends a message to a named channel.
2. Redis receives the message and identifies the channel.
3. Redis then broadcasts this message to all currently connected subscribers of that specific channel.
4. Subscribers receive the message in real-time.

It's like a radio station (channel) broadcasting to many listeners (subscribers) at once.

To publish a message, we use the PUBLISH command. It takes two arguments: the channel name and the message content.

Try running this command in your Redis CLI. It will return the number of clients that received the message.

To subscribe to a channel, we use the SUBSCRIBE command. It takes one or more channel names.

When you run this, your CLI will enter a blocking mode, waiting for messages on the notifications channel. You won't see a prompt until you stop it (e.g., Ctrl+C).

Understanding these characteristics is crucial for choosing the right Redis feature:

• Real-time: Messages are delivered instantly to active subscribers.
• Fire-and-Forget: Publishers don't know or care if messages are received.
• No Persistence: If no one is subscribed, the message is lost. Redis Pub/Sub does not store messages for later retrieval.
• Message Ordering: Messages published to a channel are delivered to subscribers in the order they were published.

You might wonder how Pub/Sub differs from other messaging patterns. The key difference is persistence.

• Pub/Sub: Ideal for real-time notifications where losing messages for offline subscribers is acceptable (e.g., chat rooms, live updates).
• Redis Streams: Provides message persistence, consumer groups, and replayability. Better for reliable queues, event sourcing, or when messages must not be lost.

We'll cover Streams in a later course!

Which of the following statements accurately describes a Redis Pub/Sub channel?

In this lesson, we've broken down the core mechanics of Redis Pub/Sub:

• Channels are the named pathways for messages.
• Publishers send messages to channels in a fire-and-forget manner.
• Subscribers listen to channels and receive messages in real-time.
• Redis Pub/Sub is not persistent; messages are lost if no subscriber is active.

Next, we'll put this into practice by broadcasting simple messages using client libraries!