Prefetch Count (QoS)
Configure the prefetch count (Quality of Service) to control how many messages a consumer receives at once. Optimize consumer efficiency and prevent individual consumers from being overwhelmed.
Prefetch Count (QoS) is a free RabbitMQ Messaging & Async Systems lesson on CoddyKit — lesson 2 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the RabbitMQ Messaging & Async Systems learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.
Control Message Flow with Prefetch
In a system with multiple consumers, how do you ensure messages are distributed fairly and no single consumer gets overwhelmed?
This lesson introduces the prefetch count, a crucial Quality of Service (QoS) setting in RabbitMQ.
The Challenge: Uneven Workload
By default, RabbitMQ dispatches messages in a round-robin fashion to available consumers. However, it doesn't wait for a consumer to finish processing a message before sending the next one.
If one consumer is slow, it might still receive many messages, while a fast consumer sits idle. This leads to an uneven workload.
Introducing Quality of Service (QoS)
RabbitMQ's Quality of Service (QoS) settings allow you to control how messages are delivered to consumers.
The most common QoS setting is the prefetchCount, which limits the number of unacknowledged messages a consumer can hold at any given time.
How Prefetch Count Works
The prefetch count tells RabbitMQ the maximum number of messages a consumer is willing to process at once.
- Once a consumer reaches its prefetch limit, RabbitMQ stops delivering new messages to it.
- It will only send more messages once the consumer acknowledges some of its current messages, bringing the unacknowledged count below the limit.
Implementing Prefetch with `basicQos`
You set the prefetch count using the channel.basicQos() method on the consumer side. The most common value for fair dispatch is 1.
This means a consumer will process one message at a time, acknowledge it, then receive the next.
// Consumer setup
Channel channel = connection.createChannel();
channel.queueDeclare("my_queue", true, false, false, null);
// Set prefetch count to 1
channel.basicQos(1, false); // prefetchCount = 1, global = false
DeliverCallback deliverCallback = (consumerTag, delivery) -> {
String message = new String(delivery.getBody(), "UTF-8");
System.out.println(" [x] Received '" + message + "'");
// Simulate work
Thread.sleep(1000);
System.out.println(" [x] Done processing '" + message + "'");
channel.basicAck(delivery.getEnvelope().getDeliveryTag(), false);
};
channel.basicConsume("my_queue", false, deliverCallback, consumerTag -> {});Sending Messages to the Queue
Let's create a simple producer that sends several messages to a queue. We'll use this with our QoS-enabled consumer.
Run this code first to populate the queue:
import com.rabbitmq.client.Channel;
import com.rabbitmq.client.Connection;
import com.rabbitmq.client.ConnectionFactory;
public class MessageProducer {
private final static String QUEUE_NAME = "my_queue";
public static void main(String[] argv) throws Exception {
ConnectionFactory factory = new ConnectionFactory();
factory.setHost("localhost"); // Assuming RabbitMQ is on localhost
try (Connection connection = factory.newConnection();
Channel channel = connection.createChannel()) {
channel.queueDeclare(QUEUE_NAME, true, false, false, null);
for (int i = 0; i < 10; i++) {
String message = "Task " + i;
channel.basicPublish("", QUEUE_NAME, null, message.getBytes("UTF-8"));
System.out.println(" [x] Sent '" + message + "'");
}
}
}
}Consumer with Prefetch in Action
Now, run this consumer code. Notice how basicQos(1) ensures that even if a consumer is slow (due to Thread.sleep), it won't hoard messages.
If you run multiple instances of this consumer, they will share the work more fairly.
import com.rabbitmq.client.Channel;
import com.rabbitmq.client.Connection;
import com.rabbitmq.client.ConnectionFactory;
import com.rabbitmq.client.DeliverCallback;
public class QosConsumer {
private final static String QUEUE_NAME = "my_queue";
public static void main(String[] argv) throws Exception {
ConnectionFactory factory = new ConnectionFactory();
factory.setHost("localhost"); // Assuming RabbitMQ is on localhost
Connection connection = factory.newConnection();
Channel channel = connection.createChannel();
channel.queueDeclare(QUEUE_NAME, true, false, false, null);
System.out.println(" [*] Waiting for messages. To exit press CTRL+C");
// Set prefetch count to 1 for fair dispatch
channel.basicQos(1, false);
DeliverCallback deliverCallback = (consumerTag, delivery) -> {
String message = new String(delivery.getBody(), "UTF-8");
System.out.println(" [x] Received '" + message + "'");
try {
// Simulate processing time
Thread.sleep(2000); // 2 seconds per message
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
} finally {
System.out.println(" [x] Done processing '" + message + "'");
channel.basicAck(delivery.getEnvelope().getDeliveryTag(), false);
}
};
channel.basicConsume(QUEUE_NAME, false, deliverCallback, consumerTag -> {});
}
}Key Advantages of Using Prefetch
Using an appropriate prefetch count offers several benefits:
- Fair Dispatch: Messages are distributed more evenly among competing consumers, preventing a single consumer from getting overloaded.
- Prevents Consumer Overload: Slower consumers won't accumulate too many messages, reducing memory consumption and potential crashes.
- Better Resource Utilization: Ensures all active consumers are working efficiently, leading to better overall throughput for the system.
`global` Flag for Channel-Wide QoS
The basicQos() method has an optional global parameter. If set to true, the prefetch count applies to all consumers on that channel.
Generally, it's safer to set global to false (the default) so the prefetch count applies per-consumer. This offers finer control and prevents unexpected behavior.
// Per-consumer QoS (recommended)
channel.basicQos(1, false);
// Channel-wide QoS (use with caution)
// channel.basicQos(1, true);Optimizing Your Prefetch Count
The ideal prefetch count depends on your application's specifics. Consider:
- Message Processing Time: If messages are processed quickly, a higher prefetch count can increase throughput.
- Consumer Resources: How much memory and CPU can each consumer dedicate to holding and processing messages?
- Network Latency: For high-latency networks, a slightly higher prefetch can reduce idle time waiting for the next message.
Experimentation is key to finding the optimal balance between throughput and fairness.
Prefetch Count Understanding
Consider two consumers, C1 and C2, both connected to the same queue. C1 processes messages in 5 seconds, C2 in 1 second. If basicQos(1) is set for both, and 10 messages are sent, what is the primary benefit?
Recap: Controlling Message Flow
Great job! You've learned about the prefetch count (QoS) in RabbitMQ.
- It limits unacknowledged messages a consumer holds.
channel.basicQos(1)is common for fair dispatch.- It prevents consumer overload and ensures even workload distribution.
Mastering prefetch count is vital for building robust and scalable messaging systems with competing consumers.
Frequently asked questions
Is the “Prefetch Count (QoS)” lesson free?
Yes — the full text of “Prefetch Count (QoS)” is free to read here on the web, and the RabbitMQ Messaging & Async Systems course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the RabbitMQ Messaging & Async Systems course, upgrade to CoddyKit PRO.
What will I learn in “Prefetch Count (QoS)”?
Configure the prefetch count (Quality of Service) to control how many messages a consumer receives at once. Optimize consumer efficiency and prevent individual consumers from being overwhelmed. You practise RabbitMQ Messaging & Async Systems with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.
Do I need any experience to start RabbitMQ Messaging & Async Systems?
No prior experience is required. RabbitMQ Messaging & Async Systems on CoddyKit is structured for beginners through advanced learners; this is — lesson 2 of 4, so you can start here or from the beginning and move at your own pace.
How long does the “Prefetch Count (QoS)” lesson take?
Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.
Can I write and run code in this RabbitMQ Messaging & Async Systems lesson?
Yes. Every RabbitMQ Messaging & Async Systems lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.