PolarDB for MySQL supports multi-zone deployment across availability zones (AZs). Compared to a single-zone deployment, a multi-zone deployment provides enhanced disaster recovery capabilities and can effectively protect your cluster from data-center-level failures. PolarDB offers several deployment options, including Dual-zone with storage standby, Dual-zone with storage and compute standby, and Three-zone. You can choose the option that best fits your use case.
Multi-zone deployment
In a multi-zone deployment, your data is distributed across multiple availability zones to ensure a higher service level agreement (SLA). The primary zone and the secondary zone each store three data replicas, for a total of six replicas. Compute nodes reside in the primary zone. The hot standby storage cluster and, in some modes, the compute nodes in the secondary zone are used for failover if the primary zone fails.
-
If you enable automatic cross-zone failover, the cluster automatically fails over from the primary to the secondary zone when a failure occurs.
-
For disaster recovery, performance drills, or to co-locate your database with ECS instances for lower latency, you can manually switch the primary and secondary zones.
The following diagrams illustrate the architecture of each deployment option.
A multi-zone deployment has the following requirements:
-
The cluster must be in a region where PolarDB has resources deployed in two or more availability zones. Multi-zone deployment is not supported in the following regions: China (Qingdao), China (Hohhot), China (Chengdu), South Korea (Seoul), Philippines (Manila), and Thailand (Bangkok).
-
The secondary zone must have sufficient compute resources.
High-availability modes
Overview of options
Single-zone
-
This mode provides database services from a single primary zone without a hot standby storage cluster, making it a low-cost option.
-
If a zone-wide failure occurs, recovery takes longer. For workloads that require cross-zone high availability, we recommend enabling a hot standby storage cluster.
Switching rules
You can switch to Dual-zone with storage standby or Dual-zone with storage and compute standby.
Use cases
-
Small websites or applications
Ideal for users who need a cost-effective cloud database to offload O&M and focus on business development.
-
Personal learning
Database beginners can use the basic series for testing and learning purposes.
-
Development and testing
Improves development efficiency by allowing you to create and release resources on demand through the rapid delivery of cloud databases.
Dual-zone with storage standby
-
Data is distributed across multiple availability zones. The primary zone and secondary zone each store a complete copy of the data, ensuring higher SLA reliability.
-
All compute nodes must be in the primary zone. The hot standby storage cluster in the secondary zone is used for failover if the primary zone fails.
Switching rules
You can only switch to the Dual-zone with storage and compute standby mode.
Billing
Data is distributed across multiple availability zones, with the primary zone and secondary zone each storing three data replicas (six replicas in total). Storage costs are higher compared to the Single-zone mode.
Use cases
This classic high-availability architecture is suitable for over 80% of use cases, including industries such as internet, IoT, retail e-commerce, logistics, and gaming.
Dual-zone with storage and compute standby
-
Data is distributed across multiple availability zones. The primary zone and secondary zone each store a complete copy of the data, ensuring higher SLA reliability.
-
The hot standby storage cluster in the secondary zone is used for failover if the primary zone fails. The number of compute nodes in the secondary zone is the same as in the primary zone. After a failover, this ensures sufficient read-only node resources to prevent service degradation.
Switching rules
You can only switch to the Dual-zone with storage standby mode.
Billing
-
Compute nodes
Additional compute nodes are added to the secondary zone that contains the hot standby storage cluster. By default, the number and specifications of these nodes match those in the primary zone, and they are billed separately.
-
Storage space
Data is distributed across multiple availability zones, with the primary zone and secondary zone each storing three data replicas (six replicas in total). Storage costs are higher compared to the Single-zone mode.
Use cases
This mode is suitable for core production databases of large and medium-sized enterprises with high-availability requirements. Examples include finance, internet new retail, automotive manufacturing, education, and large-scale enterprise ERP systems.
Three-zone (storage standby and logger)
-
This mode provides strong data consistency across multiple availability zones. Compared to semi-synchronous and asynchronous replication, its three-node architecture (one primary node, one Standby node, and one logger node) combines physical replication with the X-Paxos protocol to deliver superior disaster recovery.
-
If the primary zone cluster fails, the system automatically fails over to the secondary zone and performs cross-zone, multi-replica strong synchronous replication. This guarantees a Recovery Time Objective (RTO) of less than 60 seconds and strong data consistency (RPO = 0), providing financial-grade reliability.
Switching rules
Switching to other high-availability modes is not supported.
Billing
-
Compute nodes
Standby nodes and logger nodes are free of charge. Primary nodes and read-only nodes are billed as standard compute nodes.
-
Storage space
Data is distributed across multiple availability zones, with the primary zone and secondary zone each storing three data replicas (six replicas in total). Storage costs are higher compared to the Single-zone mode.
Use cases
This mode is suitable for core production databases of large and medium-sized enterprises with high-availability requirements. Examples include finance, internet new retail, automotive manufacturing, education, and large-scale enterprise ERP systems.
For information about the billing of compute nodes and storage space, see Billing.
Enable or switch high-availability modes
-
When you purchase a cluster, you can select the desired high-availability mode in the Network and Zone section.
NoteThe available high-availability mode options depend on the number of availability zones and resources in the cluster's region. The options shown on the purchase page are final.
-
Go to the page. Click the ID of the target cluster. On the Basic Information page, find the Distributed Storage section and click Switch to HA Mode.
-
Go to the page. Click the ID of the target cluster. Go to the page. On the Cluster Availability tab, click Switch to HA Mode.
Switching high-availability modes is subject to certain limitations. If your cluster does not support switching to the target high-availability mode, we recommend purchasing a new cluster and selecting the required mode during creation. You can then use Data Transmission Service (DTS) to migrate data from the existing cluster to the new one. For detailed instructions, see Migrate data between PolarDB for MySQL clusters.
FAQ
Check standby status
You can check this by using the PolarDB console or by making an API call.
Console
Log on to the PolarDB console. In the left-side navigation pane, click Clusters. Select the region where your cluster is located. Then, click the target cluster ID.
On the cluster details page, view the Distributed Storage section at the bottom.
API
You can call the DescribeDBClusterAttribute API to view the cluster's attributes. In the response, the HotStandbyCluster parameter indicates whether a hot standby storage cluster (and a Standby compute node) is enabled. Valid values:
StandbyClusterON: Indicates that a hot standby storage cluster is enabled. In some modes, this also indicates that a Standby node is enabled.StandbyClusterOFF: Indicates that the hot standby storage cluster is disabled.