RabbitMQ on the other hand is written in Erlang and based on the AMQP protocol. AMQP is significantly more powerful and complicated than JMS as it allows for peer to peer messaging, request/reply, and publish/subscribe models for one-to-one or one-to-many message consuming.
RabbitMQ has gained popularity in the past five years and is nowadays the most searched for queuing system.
RabbitMQ's architecture is outlined below:
Source: https://content.pivotal.io/blog/understanding-when-to-use-rabbitmq-or-apache-kafka
Scaling in RabbitMQ systems is performed by creating a cluster of RabbitMQ servers.
Clusters share data and state, which are replicated, but message queues are distinct per node. To achieve high availability we can also replicate queues in different nodes.