Docs

Message Queue for RabbitMQ

Product Overview

Product specification

The message queue for RabbitMQ is compatible with open source RabbitMQ. It offers both stand alone and cluster versions. Each version is available in several computing specifications to choose from.

Instance types

The message queue for RabbitMQ offers two types of instances: stand alone version and cluster version.

Stand alone version: deploy a single RabbitMQ broker.

Cluster version: deploy multiple RabbitMQ brokers for high-reliability messaging services.

Computing specifications

Computing specifications determine the business processing performance of RabbitMQ instances. The performance of RabbitMQ instances primarily depends on factors such as the queue count, connection count, and message accumulation.

tip

TPS peak refers to the number of messages processed per second by RabbitMQ instances, including sending messages TPS and consuming messages TPS.

When TPS peaks, queue counts, connection numbers, and other metrics approach specification limits, we recommend that you determine the need for urgent scaling based on business requirements. For example, if disk usage consistently exceeds 85% of the specification limit, scaling is recommended. You can monitor various specification metrics in real time through cloud monitoring alarms. For recommended alarm rules, refer to Recommended configuration alarm rules.

If the production link is below 200, there will be a higher upper limit of TPS performance. Please test according to your business scenario and reserve peak resources.

Normal queue

Instance types	Specification name	Recommended TPS peak	Recommended number of queues	Maximum number of connections	Maximum number of Channels	Storage space range
Stand alone version	rabbitmq.n1.x2.medium	3,000	200	2,000	6,000	100~4000 GiB
Cluster version	rabbitmq.n3.x2.small	4,500	300	1,200	3,600	300~12000 GiB
		rabbitmq.n3.x2.medium	9,000	600	3,000	9,000	300~12000 GiB
		rabbitmq.n5.x2.medium	15,000	1,000	5,000	15,000	500~20000 GiB
		rabbitmq.n7.x2.medium	21,000	1,400	7,000	21,000	700~28000 GiB
		rabbitmq.n7.x2.large	42,000	2,800	14,000	42,000	700~28000 GiB
		rabbitmq.n9.x2.large	54,000	3,600	18,000	54,000	900~36000 GiB

High queue

Instance types	Specification name	Recommended TPS peak	Recommended number of queues	Maximum number of connections	Maximum number of Channels	Storage space range
Stand alone version	rabbitmq.n1.x4.small	1500	200	1000	3000	100~4000 GiB
		rabbitmq.n1.x4.medium	3000	400	2000	6000	100~4000 GiB
		rabbitmq.n1.x4.large	6000	800	4000	12000	100~4000 GiB
Cluster version	rabbitmq.n3.x4.small	4500	600	3000	9000	300~12000 GiB
		rabbitmq.n3.x4.medium	9000	1200	6000	18000	300~12000 GiB
		rabbitmq.n3.x4.large	18000	2400	12000	36000	300~12000 GiB
		rabbitmq.n5.x4.small	7500	1000	5000	15000	500~20000 GiB
		rabbitmq.n5.x4.medium	15000	2000	10000	30000	500~20000 GiB
		rabbitmq.n5.x4.large	30000	4000	20000	60000	500~20000 GiB

Storage specifications

Storage space is mainly used to store messages. Persistent messages are saved on disk. When memory usage is high, some data will also be offloaded to disk. When selecting storage space capacity, choose appropriate storage based on business message length and replica count. For stand alone RabbitMQ instances, you can estimate the required storage space by summing the business message length and reserved disk size. For clusters with mirrored queues enabled, consider additional disk space used by redundant messages.

tip

Because some of the storage will be used for logs and metadata storage, the actual available storage will be less than the selected value.

When the remaining disk space is less than 5 GiB, RabbitMQ blocks the producer to reduce the possibility of a service crash due to disk exhaustion.

Last updated: 2026.09.21 17:32:46