We now have a need for CS software to implement peer-to-peer message push with cloud servers. Because the cs software cannot use websockt, we use golang to write a message service based on http persistent connection, but because the identification of http connection is mainly based on port, and this message service is a single point, it cannot be clustered (it is very expensive to support cluster service instead), so we intend to change the push method. After looking around the solution, we want to use rabbitmq, and implement a simple peer-to-peer push demo, locally, but we don"t know if this approach has been applied to large-scale connection scenarios. What are the conditions under which the number of consumers of RabbitMQ depends?