RabbitMQ clustering -

I created rabbit MQ cluster on single window machine with hi policy and created two DISC and two RAM nodes and 1 state node. is . I then ran the PerfTest (rabbitmq customer test utility), the result was disappointing, it was approximately 5000 meters / second. But when I took part in the same exam with a single rebitmaku node, I got good results, that is what is going wrong in 25000 meters / second, should be effective as a result, if the cluster can be run inside, but it is the opposite Anyone has to face it or know the reason behind it. Thanks

A rabbit MQ cluster with rows of mirrors will not go faster than a single node. Why? Clustering to improve reliability and fault tolerance, not to improve throughput

What is the reason? When you enable the mirror queue, RabbitMQ needs to coordinate the state between the nodes, that is, to deliver the same message more than once, or for more than one consumer, publishing, consumers and a There is no need to coordinate. All this affects coordination performance, but it is disposed of with this type of replication.

If you need decentralized replication, then you can use the Federation plugin