High-availability Edition

The High-availability Edition uses a classic primary/standby high-availability architecture. It is suitable for over 80% of use cases in industries such as the Internet, Internet of Things (IoT), retail e-commerce, logistics, and gaming.

On the Basic Information panel of an instance, the Type and Edition field displays Standard Instance (High-availability Edition).

Topology

Benefits

• High availability

  High-availability Edition instances include a standby node. Data is replicated from the primary node to the standby node using either semi-synchronous or asynchronous replication, depending on the selected parameter template. If the primary node becomes unavailable, the service automatically fails over to the standby node.

  Note

  • Semi-synchronous replication: When an application updates the primary instance, the instance transmits its logs to the standby instance. A transaction is committed once the standby instance receives the logs, without waiting for the standby to replay them. If the standby instance becomes unavailable or a network error occurs between the primary and standby instances, semi-synchronous replication falls back to asynchronous replication.

  • Asynchronous replication: When an application sends an update request (such as an insert, delete, or update), the primary instance responds immediately after completing the operation and asynchronously replicates the data to the standby instance. Therefore, an unavailable standby instance does not affect operations on the primary instance. However, if the primary instance fails, data loss may occur due to replication lag.

  The primary and standby nodes of a High-availability Edition instance can be deployed in the same or different availability zones within the same region. Deploying them in different availability zones enables cross-zone disaster recovery and improves the instance's availability.

  Note

  If the standby node fails, the primary node performs a real-time backup. Towards the end of the backup process, a global lock (FTWRL) is applied, making the primary node read-only. This process typically lasts no more than 5 seconds.

• Comprehensive features

  The High-availability Edition offers a comprehensive set of features, including auto scaling, backup and recovery, performance optimization, and read/write splitting. Its SQL Explorer feature stores up to five years of SQL execution records. This provides a detailed audit trail for all database access and helps secure your core data.

Limitations

For performance reasons, the primary and standby nodes of a High-availability Edition instance cannot span regions.

Upgrade to the High-availability Edition

Basic Edition instances do not have a standby node for hot backup. As a result, they may experience extended downtime during unexpected failures, configuration changes, or version upgrades. Use the High-availability Edition if your business requires high database availability.

Purchase an instance

FAQ

• Can I access the standby node?

  No. You can only access the primary node. The standby node is for backup only and does not serve application traffic.

• Can I downgrade a High-availability Edition instance to the Basic Edition?

  Downgrading to the Basic Edition is not currently supported. As a workaround, you can purchase a new Basic Edition instance, migrate your data to the new instance, and then release the original instance.

• How do I change the data replication mode for a High-availability Edition instance?

  See Query and modify the data replication mode.

• If the processing capacity of the standby node cannot keep up with the primary node, will the data replication mode fall back to asynchronous replication?

  Yes. For more information, see Data replication modes.