Apache Kafka vs Active MQ

Although Apache Kafka and Active MQ were designed for different purposes, there is some overlap which is worth comparison.

Apache Kafka vs Active MQ – what do they do?

As a distributed streaming platform, Kafka allows streamed data to be processed and reprocessed on disk.  Its high throughput means it is more commonly used for data streaming in real-time.

Active MQ, on the other hand, is a message broker for more general purposes.  It is able to support various messaging protocols including AMQP, STOMP, and MQTT.  It is commonly used for integration between application/services.

Explore Apache Kafka vs RabbitMQ

Apache Kafka vs Active MQ – design

Active MQ was based on JMS specification, which is a standard method used when interfacing with message brokers, whereas the design of Kafka was related to the needs of LinkedIn.  This ‘universal data pipeline’ needed a different type of message broker.  Active MQ is not able to support these data pipelines as well as Kafka.  Although Kafka does offer persistence it doesn’t come with the same guarantees as brokers based on JMS.

Apache Kafka vs Active MQ – reliability

Active MQ concerns itself with reliable and persistent messaging, and minimizing the burden on the messaging clients.  One example of this is that it is the broker’s responsibility to retain messages until consumers have processed them.  This increases the complexity and degrades the performance as the number of consumers increases.

Apache Kafka, on the other hand, uses a ‘topic’ (unified destination model) model.  In comparison to Active MQ messages are kept for a while but persistence is time-limited and the responsibility is on the consumer to consume any relevant messages before they are deleted.  This means that messages can be lost.

Apache Kafka vs Active MQ – disk space usage and capacity

The Kafka message broker can increase disk usage by up to 1000 times that of Active MQ.  Active MQ does consume a little more disk space because it keeps messages for consumers.  To offset this disk usage Active MQ allows the user to set a message lifetime, and the message will be deleted after a certain time period.

Kafka offers no guarantees to consumers who don’t pick up their messages, but it has speed and the capacity to handle millions of messages each second.

Today almost all message brokers have a similar total capacity in MB/s.  Kafka does particularly well with small messages in comparison to others, but similar at around the 75MB/s mark.

Kafka vs Active MQ

Where Kafka clusters the publishing and consumers are required to track subscriptions, Active MQ cluster the publishing of messages and also track consumer subscriptions.

  Conclusion

If you are looking for higher performance monitoring and message loss is not so critical, Kafka is ideal for your purposes.  On the other hand, a classical message broker such as Active MQ will be ideal if you care about one-time delivery and have valuable messages.

Leave a Comment