All Products
Search
Document Center

AnalyticDB:Cross-zone disaster recovery

Last Updated:Dec 15, 2025

AnalyticDB for MySQL allows you to create Enterprise Edition clusters with multi-zone deployment. Compared with single-zone clusters, multi-zone clusters provide enhanced disaster recovery capabilities and can withstand data center failures. When a failure occurs, the system automatically performs failover to ensure normal business operations.

Multi-zone deployment architecture

AnalyticDB for MySQL, based on the capabilities of regional Enterprise SSDs (ESSDs), implements cross-zone disaster recovery. The access nodes, compute nodes, and storage nodes of a multi-zone cluster are deployed in the primary zone. The data written to the cluster is automatically stored redundantly in two zones. When the primary zone fails, the entire service of AnalyticDB for MySQL automatically switches over to the secondary zone and completes the primary/secondary failover.

The following figure shows the cross-zone disaster recovery architecture of AnalyticDB for MySQL:

image

Notes

  • Only Enterprise Edition clusters and Data Lakehouse Edition support multi-zone deployment.

  • The following regions support multi-zone deployment: China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Shenzhen), China (Hong Kong), and Singapore.

  • To achieve high availability, data must be written synchronously to two separate zones. This results in a slightly higher average write latency compared to a single-zone cluster. Based on internal testing, this performance difference is typically within 5%.

Process of primary/secondary failover

When the primary zone fails, service requests of an AnalyticDB for MySQL cluster are automatically switched over to the secondary zone. This ensures a recovery point objective (RPO) of 0 and a recovery time objective (RTO) of 30 minutes or less. The following figure shows the process:

image

Impacts of primary/secondary failover

During a primary/secondary failover, data read/write operations may fail, or read/write timeout errors may occur.

Next steps