Apache Kafka vs RabbitMQ & ActiveMQ

In this clip from the webinar "Harnessing Streaming Data with Apache Kafka," Javier Perez, Connor Penhale, and Joe Carder explain how Apache Kafka is different from message brokers like ActiveMQ and RabbitMQ.

Transcript*

Connor: I think it's gotta come down to cues versus topics. So if you need guaranteed only once delivery and you don't wanna deal with an item potency channel on the side, or if everything I just said is way over your head, ActiveMQ might be a really good fit for you. Because it has a button you can press that makes sure that if you put one message on the queue, that one consumer receives it only once. Great application for transactional stuff where I don't want to have three checks deposited in my account, or three checks not deposited into my account or debited.

Javier: I mentioned at the beginning that I used to work at Red Hat and they are big contributors to ActiveMQ — not much on RabbitMQ, but ActiveMQ — and it works very closely with Java stacks, right? So doing Java development, kind of like the preferred middleware would be ActiveMQ. One of the nice things about Kafka is that it's completely independent. Your applications could be on any programming language, any framework, and then integrate with Kafka. So that's obviously another recommendation there.

Connor: The J in JMS stands for Java Messaging System, right? And that's what the standard ActiveMQ implements.

Joe: Yeah. ActiveMQ is great. It does what it does really well. It's kind of dated at this point, though, I think. And there's a big Venn diagram of ActiveMQ and Kafka, and they overlap pretty strongly, but I just think the Kafka ecosystem is a little bit more cutting edge. It's a little bit more forward, future-looking rather than the traditional JMS spec.

* lightly edited for clarity

[knful]

jms is for "java message service" not "system"