- Documentation
Message Queue for RocketMQ
User Guide
Topic management
Managing lite topics
Topic management
Managing lite topics
Managing lite topics
This topic explains how to manage lite topics.
A lite topic is a type of topic offered for Message Queue for RocketMQ 5.x instances. It is particularly useful for large-scale, fine-grained, and dynamic communication scenarios. Compared with traditional topics, lite topics offer the following benefits:
- Massive topic resources: Message Queue for RocketMQ allows you to create millions of lightweight, dynamic topics with automated lifecycle management.
- Flexible subscription model: Consumers in the same consumer group can subscribe to different sets of lite topics and dynamically add or remove subscriptions. This eliminates the need for all group members to have the same subscriptions, allowing for subscription isolation and exclusive consumption.
- Ordered messaging: Each lite topic contains only one queue. Messages are stored and consumed in strict order and can only be processed by one consumer.
- Automated lifecycle: Lite topics are created automatically when a message is sent or a subscription is made. They are also deleted automatically after they expire and no new messages are received. This eliminates the need for manual resource creation and reclamation, greatly reducing complexity.
- Fine-grained resource isolation: Lite topics support fine-grained isolation of storage and permissions. You can manage storage and permissions independently for each topic.
- Long AI session management
- In enterprise large language model (LLM) applications, a single session often takes a long time. These sessions depend heavily on context and involve multiple rounds of message exchanges and returns of intermediate results. However, when multiple sessions share one queue, messages may become blocked or processed out of order. This can cause tasks from different users to interfere with each other. To prevent this, you can map each AI session to an independent message stream and associate it with a dedicated lite topic. The system processes these message streams in the correct order, preserving the order and isolating the context for long sessions. Specifically:
- Session isolation: Each session is carried out independently to minimize interference.
- Strict sequential processing: The single-queue model is the default option and is ideal for AI session pipelines that require context to be preserved in a specific order.
- Massive dynamic sessions: A lite topic is capable of handling millions of AI sessions without requiring pre-maintenance of traditional queue resources for each session.
- Multi-agent task orchestration
- Enterprise AI applications often require multiple agents to divide labor in order to complete a single user request. This process includes task decomposition, task processing, and result aggregation. Such scenarios require higher levels of message isolation, asynchronous decoupling, and resource elasticity. Lite topics are ideal for distributing tasks and collecting results between a primary agent (supervisor) and secondary agents (sub-agents). The system can dynamically create a lite topic for each task and automatically release them once complete. This supports asynchronous decoupling and parallel processing in multi-agent collaboration scenarios. Specifically:
- Task isolation: Different agents and tasks do not interfere with each other, which prevents task blocking.
- Dynamic creation and automatic release: Lite topics are ideal for temporary, on-demand communication channels for agent tasks.
- Asynchronous parallel processing: The main process can quickly dispatch tasks, and sub-tasks process them independently and return results.
- Ordered and isolated results: Each task has an independent lite topic, which facilitates task distribution and result collection.
- Lite topics are currently in closed beta. Please contact your account manager for access.
- Lite topics are only available with RocketMQ 5.x instances.
When creating a RocketMQ 5.x instance, add the lite topic feature tag. Set the tag key to mq:sys:version-capability and the tag value to lite-topic. For instructions on creating an instance, see Creating instances. - On the Instance List page, click the instance you want to manage.
- On the Topic Management tab, click Create Topic.
- In the Create Topic panel, configure the following parameters.
Viewing lite topic details
You have created a lite topic.
- On the Instance List page, click the instance you want to check.
- On the Topic Management tab, click the lite topic you want to check.
- On the topic details page, view the topic's Basic information, Consumer group information, Key permission, and Cloud monitoring information.
- Consumer group information
- This section displays information about the consumer groups associated with the lite topic. For more information, see Creating groups.
Editing the auto-delete settings for a lite topic
You have created a lite topic.
- On the Instance List page, click the instance you want to manage.
- Open the edit dialog box in one of the following ways:
- Option 1: On the Topic Management tab, click Edit in the Operation column of the lite topic you want to manage.
- On the Topic Management tab, click the lite topic you want to manage.
- In the Basic information section, click Edit under Automatically delete LiteTopic.
- In the Edit dialog box, enable or disable Auto delete. If you enable this option, specify an Expiration time.
- Save your settings by clicking Save.
- You have created a lite topic.
- You have deleted all groups associated with the lite topic in the LITE_SELECTIVE consumption mode.
- On the Instance List page, click the instance you want to manage.
- On the Topic Management tab, click Delete in the Operation column of the lite topic you want to remove.
- In the box that appears, verify the name of the topic and the associated instance ID.
tip
If the system detects any groups in the LITE_SELECTIVE consumption mode, select them and click Batch deletion to remove the association.
Last updated: 2026.09.23 10:34:37