Reliability Functions
Cluster HA
A Kafka instance has multiple replicas for instance-level DR. If the leader replica is faulty, another leader is quickly elected to ensure uninterrupted Kafka services.
In a RabbitMQ cluster, data is replicated to all nodes through mirrored queues, preventing service interruption in case of a node breakdown.
A RocketMQ instance has one master and two slaves. To ensure strong data consistency, the slaves synchronize data from the master. When a node fault occurs, the master/slave roles are automatically switched by using the Raft protocol.
Cross-AZ DR
Kafka, RabbitMQ, and RocketMQ instances can be deployed across AZs. At least three AZs are required. If one AZ is faulty, the instance still provides services.
Monitoring
Kafka: Use Cloud Eye to monitor and report alarms for Kafka instances, instance nodes, instance topics, instance partitions, consumer groups of instance partitions, consumer groups of instance queues, and consumer groups of instances.
For details, see Kafka Metrics.
RabbitMQ: Use Cloud Eye to monitor and report alarms for RabbitMQ instances, instance nodes, and instance queues.
For details, see RabbitMQ Metrics.
RocketMQ: Use Cloud Eye to monitor and report alarms for RocketMQ instances, instance nodes, instance queues, consumer groups of instances, consumer groups of instance queues, and dead letter queues of instances.
For details, see DMS for RocketMQ Metrics.
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot
