We have been working with RabbitMQ a long time now, and we have probably seen way more configuration mistakes than anybody else. We know how to configure for optimal performance and how to get the most stable cluster. The CEO, Carl Hörberg, shared his knowledge in a free webinar hosted by Erlang Solutions.
A link to the requested webinar can be found here.
We have also written a guide on the same subject, where we talk about dos and don'ts mixed with best practice for two different usage categories; high availability and high performance (high throughput). Among other things, we discuss queue size, common mistakes, lazy queues, prefetch values, connections and channels, HiPE and number of nodes in a cluster. All generally best practice rules, based on the experience we have gained while working with RabbitMQ.
A link to the RabbitMQ Best Practice guide can be found here.