Multi-node High Availability
Last updated: 2025-08-14 11:39:56
Database High Availability (HA) ensures that the system can continue providing services when some nodes fail via a redundancy design and the automatic failure recovery mechanism. Read-only analysis engine instances achieve high availability by adopting the multi-node deployment mode. Database instances are distributed across multiple physical nodes. In addition, data synchronization, failure detection, traffic switching, and other mechanisms are used together to ensure business continuity. This document describes the principles of achieving high availability for read-only analysis engine instances, as well as relevant operations and management.
Principles of Achieving High Availability for Read-Only Analysis Engine Instances
Redundancy Architecture
One read-only analysis engine instance supports deploying 1 to 3 nodes. When the number of nodes is 2 or 3, they can form a peering cluster. If any node of the read-only analysis engine instance fails, the system can automatically forward SQL traffic to other "live" nodes that are running normally to avoid business unavailability caused by a single point of failure.
Fault Detection and Failover
Heartbeat mechanism: The node "liveness" status of a read-only analysis engine instance is detected through periodic heartbeats. If a node does not respond within a certain time period (detection interval: 3 seconds), fault determination is triggered.
Automatic failover: When a node fault is detected through periodic heartbeats, the system will perform two consecutive "liveness" detections with no interval (duration: about 10 seconds). If the node still fails, the system will remove the faulty node from the read-only analysis engine instance within 1 to 5 seconds. Then, it waits for the faulty node to recover or be automatically rebuilt and adds the node to the read-only analysis engine instance again.
Load Balancing
Load balancing is adopted for nodes that provide services of read-only analysis engine instances. All nodes can normally handle business traffic, which improves concurrent execution performance.
Billing Instructions
Read-only analysis engine instances are divided into single-node and multi-node instances according to the number of nodes. The price of multi-node instances differs from that of single-node instances. For more information, see Billing Overview.
Adjusting the Number of Nodes
Note:
Read-only analysis engine instances currently adopt the method of pulling data from the source instance of TencentDB for MySQL by default to rebuild nodes. Therefore, it is recommended that you perform node addition operations during off-peak hours.
1. Log in to the TencentDB for MySQL console.
2. Find the target read-only analysis engine instance in the instance list and click Manage in the Operation column to enter the instance details page.
3. Choose Instance Details > Instance Nodes and click Modify Node.
4. Add or delete nodes as needed on the page that appears, and then click Submit.
Add a node: To add a node, click Add Node. A node is then added to the current read-only analysis engine instance. Each read-only analysis engine instance supports up to 3 nodes.
Delete: To delete a node, click Delete. If you delete a node by mistake, click Restore before clicking Submit.
Note:
The specification and AZ of the node are the same as those of the read-only analysis engine instance.
At least 1 node should be retained for each read-only analysis engine instance. Deleting all instances is not supported.
5. After the operation is completed, you can return to the instance details page. The node can be used normally when the displayed status is "Running".
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No
Feedback