การจัดการกลุ่มผู้บริโภค

Welcome! Today, we'll explore Consumer Groups in Kafka. These are a core concept for scaling message consumption and ensuring fault tolerance.

Think of a consumer group as a team of consumers working together to process messages from one or more topics. Each message from a topic's partitions is delivered to only one consumer within that specific group.

Consumer groups solve two major challenges:

• Scaling: By distributing partitions across multiple consumers, you can process messages much faster, increasing your application's throughput.
• Fault Tolerance: If a consumer fails or leaves the group, Kafka automatically reassigns its partitions to other active consumers in the same group. This ensures continuous message processing.

Ultimately, a message is processed exactly once per consumer group, even with multiple consumers.

To understand groups, remember that Kafka topics are divided into partitions. These partitions are the unit of parallelism.

• Within a consumer group, each partition is assigned to at most one consumer.
• If you have more consumers than partitions in a group, some consumers will be idle.
• If you have fewer consumers than partitions, some consumers will handle multiple partitions.

This assignment strategy ensures ordered processing within each partition while allowing parallel processing across partitions.

Kafka is smart! When a consumer joins or leaves a group (e.g., an application starts, stops, or crashes), Kafka automatically triggers a rebalance.

During a rebalance, partitions are dynamically re-assigned among the active consumers in the group. This ensures all partitions continue to be consumed and adapts to changes in your application's scaling needs.

Rebalancing is an automatic process that guarantees high availability and adapts to fluctuating workloads.

In Spring Kafka, you define which consumer group your listener belongs to using the group.id property.

This unique string identifies your consumer group to Kafka. All consumers (whether in the same application instance or different instances) that share the exact same group.id are considered part of the same group.

You typically configure this in your application.properties or directly in the @KafkaListener annotation.

Here's a basic Spring Boot Kafka listener configured with a specific group.id. This tells Kafka that this listener is part of the 'my-first-group' consumer group.

Try running this example and observe the output!

To achieve parallel processing and higher throughput for a topic, you can run multiple instances of your application, all configured with the same group.id.

Kafka will then distribute the topic's partitions across these running instances. Each instance will process a subset of the partitions independently, effectively scaling out your message consumption.

This is the primary way to handle high-volume topics efficiently.

You can also define multiple @KafkaListener methods within the same Spring Boot application instance, all belonging to the same consumer group.

Spring Kafka manages these as separate consumers within that group. If the topic has enough partitions, these listeners can process messages from different partitions in parallel within a single application.

The id attribute helps differentiate listener beans, especially when multiple listeners are defined for the same topic and group.

What if different applications need to process the same messages from a topic, but for different purposes?

This is where multiple consumer groups come in handy. You can have several distinct consumer groups, each with its own unique group.id, consuming from the same Kafka topic.

Each group will receive a full copy of all messages published to that topic. This allows for independent processing without affecting other groups.

You've learned about the power of Kafka Consumer Groups!

• They are crucial for scaling message consumption and building fault-tolerant applications.
• Messages from a topic's partitions are distributed among consumers in a group, processed exactly once per group.
• Kafka handles partition assignment and rebalancing automatically as consumers join or leave.
• Using different group.id values enables multiple applications to independently consume the same stream of messages.

Mastering consumer groups is key to building robust Kafka applications.