รูปแบบคอนซูเมอร์แข่งขันกัน

Welcome! In distributed systems, you often need to process many tasks efficiently. The Competing Consumers pattern is a powerful way to achieve this.

It allows you to scale your message processing capacity by simply adding more workers.

Imagine a single queue of tasks. Instead of one worker taking all tasks, multiple workers (consumers) listen to this same queue.

• Each message is delivered to only one of the competing consumers.
• Consumers "compete" to receive the next available message.
• This distributes the workload automatically.

The Competing Consumers pattern offers several advantages:

• Scalability: Easily increase processing power by adding more consumer instances.
• Reliability: If one consumer fails, others can pick up its share of messages.
• Load Balancing: Messages are spread across available consumers, balancing the workload.
• Decoupling: Producers don't need to know how many consumers there are or where they are.

Let's set up a basic producer that sends messages (tasks) to a queue named task_queue. Each message will be a simple string.

Run this code to send a few messages:

Now, let's create our first consumer. It will connect to task_queue and start processing messages. Each message will be acknowledged after simulating work.

Run this consumer in a terminal:

To truly see the competing consumers pattern in action, open a new terminal window and run the exact same TaskConsumer code again.

You'll now have two consumer instances listening to the task_queue. Run the producer code (from Scene 4) once more. Observe how messages are now distributed between both consumer instances, demonstrating how they compete for messages and share the workload!

By default, RabbitMQ uses a round-robin dispatching mechanism when multiple consumers are connected to the same queue. This means messages are sent to consumers sequentially.

• Consumer 1 gets message 1.
• Consumer 2 gets message 2.
• Consumer 1 gets message 3, and so on.

The basicQos(1) setting in our consumer code is crucial here. It tells RabbitMQ not to send more than one unacknowledged message to a consumer at a time, ensuring fair dispatch even if consumers process at different speeds.

The Competing Consumers pattern is ideal for scenarios like:

• Image processing: Multiple workers resizing images from a queue.
• Email sending: Sending bulk emails without overwhelming a single service.
• Log processing: Analyzing large volumes of logs in parallel.
• Background jobs: Any task that can be processed independently by multiple workers.

When using competing consumers, keep these in mind:

• Message Ordering: If strict message order is critical, this pattern might not be suitable directly, as messages can be processed out of order by different consumers.
• Idempotency: Consumers should ideally be idempotent. This means processing the same message multiple times should have the same effect as processing it once. This is vital for fault tolerance and retries.

Great job! You've learned about the Competing Consumers pattern:

• It enables multiple consumers to process messages from a single queue.
• It's excellent for scaling and load balancing message processing.
• RabbitMQ's default round-robin dispatch and QoS settings facilitate fair distribution.
• Consider idempotency and potential out-of-order processing for specific use cases.

Next, we'll dive deeper into optimizing consumer efficiency with prefetch counts!