As your workload grows or shrinks, you can adjust the specifications of an ApsaraDB RDS for PostgreSQL instance—RDS edition, product type, storage type, instance type, and storage capacity—without changing the instance ID or endpoints.
Prerequisites
Before you begin, make sure that:
The instance billing method is subscription or pay-as-you-go
The instance is in the Running state
Your Alibaba Cloud account has no overdue renewal orders
To change the specifications of a Serverless ApsaraDB RDS for PostgreSQL instance, see Serverless ApsaraDB RDS for PostgreSQL instances.
What you can change
During a specification change, workloads switch over to a new instance. Depending on the change type, a transient connection or instance restart may occur—typically lasting a few minutes. Most operations related to databases, accounts, and network settings are unavailable during the switchover. Schedule specification changes within your maintenance window to minimize impact.
The following table lists what you can change and the business impact of each change.
| Configuration item | What you can do | Impact |
|---|---|---|
| RDS edition | Upgrade from Basic Edition to High-availability Edition. See Upgrade from Basic Edition to High-availability Edition. | Transient connection |
| Product type | Switch between YiTian and standard. See Product types. | Transient connection |
| Storage type | See Storage type changes below. | Depends on change type—see below |
| Instance type | Change to a different instance type. See Instance types for primary instances. | Upgrade: transient connection. Downgrade: primary instance restarts, then transient connection. |
| Storage capacity | Expand or reduce storage. See Storage capacity changes below. | Depends on storage type and direction—see below |
Storage type changes
For instances using cloud disks:
Upgrade from standard SSD to Enterprise SSD (ESSD): ESSD PL1, ESSD PL2, or ESSD PL3. Downgrading from ESSD to standard SSD is not supported.
Switch between ESSD PL1, ESSD PL2, and ESSD PL3 in either direction.
Upgrade from ESSD to Premium ESSD. Downgrading from Premium ESSD to ESSD is not supported.
For instances using Premium Local SSDs: Storage type changes are not supported.
Business impact:
Switching between ESSD performance levels (PLs), or upgrading from ESSD to Premium ESSD: no business interruption.
Upgrading from standard SSD to ESSD: transient interruption.
For a comparison of storage type performance, see Storage types.
Storage capacity changes
Expansion: Expand at a step size of 5 GB. The new capacity cannot exceed the maximum supported by the instance type. See Primary instance types.
Reduction (cloud disks only): Reduce at a step size of 5 GB, down to the larger of the instance type minimum or the result of this formula:
Minimum reducible capacity = min(Used storage × 1.3, Used storage + 400 GB)If your instance was created before October 10, 2022 and uses an original architecture, you must update the minor engine version to the latest version before reducing the storage capacity. See Update the minor engine version.
For step-by-step instructions, see Reduce storage capacity of an instance using cloud disks.
Business impact:
Cloud disk expansion: no transient connections.
Cloud disk reduction: transient connections occur.
Premium Local SSD (expansion only): transient connection. If the host has enough storage resources, no data migration occurs. If not, data migrates to a new host.
Product type change constraints
Before switching between YiTian and standard, check the following:
If the target product type is unavailable, no resources exist in the current zone. Migrate the instance to a zone that supports the target product type, then retry. See Migrate across zones in the same region.
The minor engine version after the change must be the same as or later than the current minor engine version. If it would be earlier, the change fails.
YiTian instances do not support the plv8 and rdkit extensions. When switching from standard to YiTian, make sure these extensions are not in use. See Extensions supported by ApsaraDB RDS for PostgreSQL.
Instance type change constraints
On High-availability Edition, you can change a general-purpose instance type to a dedicated instance type with more than 16 CPU cores, or back to a general-purpose type.
Some general-purpose instance types being phased out cannot be changed. See EOS notice: general-purpose instance types.
Changing to a Basic Edition instance type with YiTian architecture is not supported.
When changing the instance type, kernel parameters on read-only instances must match those on the primary instance. If kernel parameters on a read-only instance need updating, that instance restarts, causing a transient connection.
Usage notes
After the change:
The instance ID and endpoints remain unchanged—no connection string updates needed.
No manual restart is required.
RDS Basic Edition: Basic Edition has no secondary hot standby. If the instance exits unexpectedly, or during a specification change or engine version upgrade, the database may be unavailable for an extended period. For high-availability requirements, use High-availability Edition instead.
Instances with read-only instances attached:
When changing storage capacity, each read-only instance's storage must be greater than or equal to the primary instance's storage. Follow this order:
Expansion: Expand read-only instances first, then expand the primary instance.
Reduction: Reduce the primary instance first, then reduce the read-only instances.
When changing the instance type, follow this order:
Upgrade: Upgrade read-only instances first, then upgrade the primary instance.
Downgrade: Downgrade the primary instance first, then downgrade the read-only instances.
For instances using Premium Local SSDs, each read-only instance's specs must be greater than or equal to the primary instance's specs. For instances using cloud disks, each read-only instance's specs must be greater than or equal to half the primary instance's specs.
vSwitch IP address requirement:
When upgrading the instance type, the vSwitch must have at least two available IP addresses. Check available IP addresses in the Virtual Private Cloud (VPC) console. If the vSwitch doesn't have enough addresses, switch to a different vSwitch first. See Change the vSwitch.
Storage expansion during backup:
If the instance is being backed up, start the expansion after the backup completes. Schedule expansion outside the backup window.
Auto storage expansion:
Instead of manually adjusting storage capacity, enable the auto storage expansion feature. When remaining storage drops below a configured threshold, the system expands storage automatically. See Configure automatic storage expansion.
Change instance specifications
Go to the Instances page. In the top navigation bar, select the region where the instance resides. Find the instance and click its ID.
In the Configuration Information section of the Basic Information page, click Change Specifications.
(Subscription instances only) In the dialog box, select Upgrade or Downgrade, then click Next step. After you submit the order, the system syncs the instance data to a new instance and switches the ID and endpoints over. Both the Upgrade and Downgrade methods take effect immediately after you submit the order, and are supported for both subscription and pay-as-you-go instances.
WarningIf you select Downgrade, the primary instance restarts during the switchover. Schedule the change within the maintenance window.
Select the new specifications. For details on each configuration item, see What you can change.
Set Switching Time:
Execute Immediately: The system applies the change and switches workloads as soon as data migration finishes.
Switch Within Maintenance Window: The workload switchover happens during your configured maintenance window. Use this option when you want to control when the transient connection or instance restart occurs. See Set the maintenance window.
- If you only expand cloud disk storage capacity or switch between ESSD PL levels, workloads are not affected and the system switches immediately. Switch Within Maintenance Window is not needed. - For instances using Premium Local SSDs: if the host has enough storage resources, the system switches immediately with no impact. If not, data migrates to a new instance and the switchover happens during the maintenance window.
On the Change Specifications page, review the changes, read and accept the Terms of Service, click Confirm Order, and complete the payment.
Warning- Specification change orders cannot be cancelled after submission. Evaluate your requirements carefully before proceeding. - Do not run DDL operations from the time you submit the order until the change completes.
Billing
For pricing details, see Billing rules for configuration changes.
FAQ
Does expanding only storage capacity require migrating instance data?
It depends on the storage type:
Cloud disks: no data migration needed.
Premium Local SSDs: migration depends on host resources. If the host has enough storage, data stays in place. If not, the system migrates data to a host with enough resources.
How do I reduce storage capacity?
Cloud disks: see Reduce storage capacity of an instance using cloud disks.
Premium Local SSDs: use the major engine version upgrade to migrate the instance to one that uses cloud disks, then reduce capacity during the upgrade. See Upgrade the major engine version.
For instances created before October 10, 2022, using an original architecture, update the minor engine version before reducing capacity. See Update the minor engine version.
How do I change storage type from cloud disk to Premium Local SSD?
See How do I change the storage type from cloud disk to local disk?
If I upgrade CPU, memory, and storage at the same time, how long does the transient connection last?
The duration is the same as upgrading each component separately—typically a few minutes. During the switchover, most operations related to databases, accounts, and network settings are unavailable. Schedule the change within your maintenance window.
If I separately change the specs of a read-only instance, does it affect the primary instance?
In most cases, separately changing a read-only instance's specs does not affect the primary instance. However, follow the ordering requirements in Usage notes when both instances need spec changes.
API reference
| API | Description |
|---|---|
| ModifyDBInstanceSpec | Changes the configurations of an ApsaraDB RDS instance |