This topic describes the architecture of ApsaraMQ for Kafka and the publish-subscribe model.
ApsaraMQ for Kafka architecture
An ApsaraMQ for Kafka cluster consists of producers, brokers, ZooKeeper, and groups, as shown in ApsaraMQ for Kafka architecture.
Figure 1. ApsaraMQ for Kafka architecture
Producer
A producer pushes messages to ApsaraMQ for Kafka brokers. The messages sent can be page view information, server logs, and information related to system resources such as CPU utilization and memory usage.
Kafka Broker
A broker is a server used to store messages. Brokers can be scaled out. The more the brokers, the higher the throughput of the ApsaraMQ for Kafka cluster.
Group
A group pulls messages from an ApsaraMQ for Kafka broker to subscribe to topics and consume messages.
Zookeeper
Zookeeper manages the cluster configuration, elects the leader partition, and balances the load when a group changes.
Publish-subscribe model of ApsaraMQ for Kafka
ApsaraMQ for Kafka uses the publish-subscribe model, as shown in Publish-subscribe model of Message Queue for Apache Kafka.
Figure 2. Publish-subscribe model of Message Queue for Apache Kafka
Groups and topics are in the many-to-many relationship. One group can subscribe to multiple topics, and one topic can be subscribed to by multiple groups.
However, the messages of a topic to which a group is subscribed can be consumed by only one consumer in the group.