Change the instance type, public bandwidth, or disk billing method of an ECS instance to match your workload.
Change the instance type
An instance type is a predefined combination of vCPUs and memory. You cannot change vCPUs or memory individually.
Before you change the instance type, check the compatible instance types on the configuration change page.
The available method depends on the billing method of the instance.
|
Instance billing method |
Operation period |
When it takes effect |
Related operations |
|
Subscription instance |
Before expiration |
Effective after instance restart |
|
|
Within 15 days before expiration |
Takes effect if the instance is restarted within 7 days of the new billing cycle. |
||
|
After expiration but before release |
Effective after instance restart |
||
|
Pay-as-you-go instance |
Not applicable |
Effective after instance restart |
Change the billing method for network usage
The available method depends on the public IP address type.
|
Public IP address type |
Effective time |
Related operations |
|
System-assigned public IP address |
Immediately |
|
|
Elastic IP address (EIP) |
Immediately |
Modify the maximum public bandwidth
The available method depends on the public IP address type and the billing method of the instance.
-
If you change the public bandwidth from a non-zero value to 0 Mbit/s, the result depends on the network type of the instance.
-
Virtual Private Cloud (VPC): The public IP address is immediately released.
-
Classic network: The instance loses Internet access, but the public IP address is retained.
-
-
If you change the public bandwidth from 0 Mbit/s to a non-zero value, the system assigns a public IP address to the instance.
For a classic network instance, the first time you change the public bandwidth from 0 Mbit/s to a non-zero value, you must restart the instance in the ECS console or by calling RebootInstance for the change to take effect.
|
Public IP address type |
Scope |
Effective time |
References |
|
Static public IP address |
Modify the bandwidth of a subscription instance |
Takes effect immediately |
|
|
Modify the bandwidth of a subscription instance upon renewal |
Takes effect in the next billing cycle |
||
|
Temporarily upgrade the bandwidth of a subscription instance |
Takes effect during the specified period |
||
|
Modify the bandwidth of a pay-as-you-go instance |
Takes effect immediately |
||
|
Elastic IP address (EIP) |
Upgrade or downgrade the bandwidth of an EIP |
Takes effect immediately |
Change the billing methods of data disks
Only subscription instances support changing the billing method of data disks, because pay-as-you-go instances accept only pay-as-you-go data disks.
|
Change time |
Effective time |
Related operations |
|
Before the instance expires |
Immediately |
|
|
Within 15 days before expiration, or after expiration but before release |
Immediately |
FAQ
How are the fees for upgrading an ECS instance calculated?
-
For subscription instances, the upgrade fee is the price difference between the new and original configurations for the remaining subscription period.
-
For pay-as-you-go instances, billing starts at the new instance type rate after the upgrade.
The actual fee is displayed on the upgrade page. You can also check the Expenses and Costs page for details.
Does upgrading or downgrading an ECS instance affect business operations?
It might. The platform guarantees only that the instance can start with the new resources. Before you change the configuration, verify that the target CPU, memory, network, and storage performance meet your workload requirements to avoid startup failures.
What impact does upgrading or downgrading an ECS instance have on the services on the instance?
Upgrading or downgrading an instance causes a brief service interruption. Perform the operation during off-peak hours and back up your data in advance.
-
Pay-as-you-go instances: Stop the instance before changing its configuration.
-
Subscription instances: Restart the instance for the new configuration to take effect.
If any of the following conditions apply, create snapshots or images before the operation:
-
Stateful services: Perform a primary/secondary switchover or schedule a maintenance window in advance.
-
Custom images: Verify the compatibility between the target instance type and the current operating system. See Check OS compatibility.
-
Cross-generation upgrades (for example, from generation 7 to 8): NVMe disk device names may change. Check for conflicts in fstab or startup scripts. See Device names of NVMe disks.
Why is the estimated refund for a downgrade CNY 0?
When you downgrade an instance, the estimated refund may be 0. This can occur if you purchased the instance at a discount, because the cost of the downgraded configuration is calculated based on the original price. If the price difference is 0 CNY or less, the refund is 0. See Refund Rules.
Can I cancel an upgrade order to restore the original configuration?
No. After an upgrade order takes effect, it cannot be canceled. To restore the original configurations, downgrade the instance. You are charged for the downgrade.
See Instance FAQ.