Cross-region disaster recovery for ApsaraDB RDS for PostgreSQL requires linking instances across regions into a unified replication topology. Global Active Database (GAD) instance groups provide that topology — a logical grouping that connects a primary instance to a secondary instance in another region and manages asynchronous replication between them.
Prerequisites
Before you begin, make sure that the RDS for PostgreSQL instance you plan to use as the primary node meets the following requirements:
| Requirement | Supported values |
|---|---|
| Billing method | Subscription or pay-as-you-go |
| RDS edition | Basic Edition or High-availability Edition |
| Engine version | PostgreSQL 11 or higher |
| Instance type — Dedicated | 4 cores and 8 GB of memory or higher |
| Instance type — General-purpose | 2 cores and 4 GB of memory or higher |
| Region | China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Shenzhen), China (Hohhot), China (Heyuan), or China (Chengdu) |
If this is your first time configuring GAD, authorize Data Transmission Service (DTS) to access cloud resources before proceeding.
Topology constraints
Instance exclusivity: An RDS for PostgreSQL instance can belong to only one GAD instance group, whether it serves as the primary or secondary node.
Group structure: Each GAD instance group supports a strict 1:1 topology — one primary node and one secondary node.
Billing
GAD costs come from two sources: the underlying RDS instances and the DTS synchronization link. Standard subscription or pay-as-you-go rates apply to both primary and secondary nodes, with no additional surcharges for incorporating existing instances. For details, see Billing.
Create an instance group
Two entry points are available:
From the Global Active Database page — Configure all group parameters before selecting an instance.
From the Basic Information page — Start from a specific instance and let the system provision the group automatically.
Both paths produce the same result: a provisioned GAD instance group with the selected instance as the primary node.
From the Global Active Database page
Go to the Global Active Database page and click Create Instance Group.
On the Create Global Active Instance Group page, configure the following parameters:
Parameter Description Instance Group Name A unique name for the group. Must be 2–126 characters, start with a letter or Chinese character, and may contain digits, underscores (_), and hyphens (-). Database Instance Type Select PostgreSQL. Instance Group Type Select Disaster Recovery. This mode configures asynchronous, one-way replication from the primary to the secondary node for cross-region failover and recovery drills. Region of Primary Instance The region hosting the primary instance. Defaults to the current console region. To switch regions, use the global region selector at the top of the page. Primary Instance Select the existing RDS instance to serve as the primary node. If no eligible instance is available, click Create Primary Instance to purchase an RDS for PostgreSQL instance. Click OK.
The instance group is provisioned and listed in the console. Add a secondary instance to complete the disaster recovery topology.
From the Basic Information page
Go to the RDS Instances page, select a region, and click the ID of the RDS for PostgreSQL instance to use as the primary node.
On the Basic Information page, find the Instance Distribution section and click Add next to Disaster Recovery Instance. The system provisions a GAD instance group using the current instance as the primary node and redirects you to the Add Secondary Instance page.
If the instance is already the primary node in an active GAD group, the system reuses the existing group. No new group is created.
On the Add Secondary Instance page, follow the instructions in Method 1: Create a new instance and add it as a secondary instance.
To add an existing instance as the secondary, close this page and follow Method 2: Add an existing instance as a secondary instance instead.
Delete an instance group
Remove the secondary node before deleting the instance group. A group that still contains a secondary node cannot be deleted. If the Delete button is grayed out, the group still has a secondary instance — remove it first.
Deleting the instance group and releasing the primary instance are separate operations. Deleting the group dissolves the GAD topology and removes the logical grouping — the primary instance remains active with its data intact. To release or unsubscribe from the primary instance, you must delete the instance group first. Once released, the instance follows standard lifecycle policies (e.g., moving to the recycle bin).
If the primary instance is suspended due to overdue payments or subscription expiration, the GAD group is automatically dissolved and the DTS synchronization link is severed. The secondary node decouples from the group and promotes itself to a standalone read/write instance.
Go to the Global Active Database page.
Find the instance group and click Delete in the Actions column.
Read the notes in the dialog box, then click OK to confirm deletion.