Docs

Volcengine Kubernetes Engine

Intelligent Horizontal Pod Autoscaler (IHPA)

Creating IHPA

Intelligent Horizontal Pod Autoscaler (IHPA) predicts scaling targets based on data, provides fine-grained management on the number of workload replicas, and reduces the impact of native HPA scaling delay on business operations.

tip

[Beta·Apply for Trial]: This feature is currently at the beta stage. Submit an application if you need to use it.

Prerequisites

A cluster has been created, and the cluster version meets the following requirements:

Clusters of v1.24 need to be upgraded to v1.24.15-vke.13 or later.

Clusters of v1.26 need to be upgraded to v1.26.10-vke.13 or later.

Clusters of v1.28 need to be upgraded to v1.28.3-vke.15 or later.

tip

If the version of your cluster does not meet the requirements, you can upgrade the cluster. For details, see Upgrading clusters.

Cloud native observability has been enabled. For details, see Enabling observability.

Procedure

Step 1: Install an add-on

tip

If the katalyst add-on is installed in your cluster, you need to upgrade the add-on to v0.5.6-vke.6 or later. For details, see Upgrading add-ons.

In the left navigation bar, click Cluster.

On the cluster list page, click the cluster you want to configure.

On the cluster management page, select Addons in the left navigation bar.

Click the Scheduling tab. Move the pointer over the katalyst add-on card and then click Deploy.

Select IHPA.

Click Confirm to install the add-on.

Step 2: Create a scaling policy

In the left navigation bar on the cluster management page, click Autoscaling > IHPA. On the page that appears, click Create IHPA to create a scaling rule in the console.

Basic information

Parameter	Description
Name	Specify the name of the IHPA, which must be unique within the cluster.
Prediction model	Configure the prediction mode of the IHPA. Valid values: Preview: only previews workload scaling based on the configured IHPA policy. The number of workload replicas is not increased or decreased. Automatic: Workload autoscaling can be performed based on the configured IHPA policy.
Namespace	Select the namespace to which the scaling policy belongs. We recommend that you select a custom namespace or the system default namespace named default. For details about namespaces, see Namespace overview.
Label	It corresponds to the label in Kubernetes and is used to identify the scaling policy. Click Add new and then enter a key-value pair that complies with the on-screen requirements.
Annotation	It corresponds to the annotation in Kubernetes, which is used to attach non-identifying metadata to the scaling policy for efficient metadata retrieval. Annotations can contain characters that are not permitted in labels. Click Add annotation and then enter a key-value pair that complies with the on-screen requirements.

Scaling policy

Parameters	Description
Scalable object	Specify a workload to be scaled. Select the type of workload and the specific workload from the drop-down lists. For workload types, you can select Deployment or StatefulSet.
Monitoring metrics	Click Add monitoring metrics to configure the metrics for workload autoscaling. The system determines whether scaling is required based on the metrics. The following metrics are available: CPU usage: Specify the CPU utilization as a metric, which is the proportion of the total CPU usage of all pod instances to the total CPU requests of all pod instances. When CPU utilization reaches this value, scaling is triggered. Memory usage: Specify the memory utilization as a metric, which is the proportion of the total memory usage of all pod instances to the total memory requests of all pod instances. When memory utilization reaches this value, scaling is triggered.
Minimum number of instances	Specify the minimum number of instances during scaling. When the policy is triggered, workload instances will scale within the configured minimum and maximum instance limits.
Maximum number of instances	Specify the maximum number of instances during scaling. When the policy is triggered, workload instances will scale within the configured minimum and maximum instance limits.
Prediction percentiles	A predictive quantile that indicates the probability of the actual metric value not reaching the specified target value. A larger value indicates a more conservative scaling policy. The value needs to be a positive integer. Valid values: 1 to 100. Recommended values: 90 to 99.
Cold start time	The time required for a pod to reach the Ready state.

Scaling rule

Parameters	Description
Scaling policy	Select Scale-out policy and Scale-in policy to configure the workload scale-in and scale-out behavior.
Silence time	Specify the silence time of the scale-out policy or scale-in policy. The value is an integer. Unit: seconds. After the system completes scaling, the scaling policy will not be executed again during the silence time. This prevents frequent changes in the number of workload replicas that lead to system instability.
Policy type	Specify the policy type. Valid values: Maximum value: executes the policy that has the most impact on pod count. Minimum value: executes the policy that has the least impact on pod count. Disable: disables the scale-out policy or scale-in policy.
Policy	Configure specific scaling policy operations. Valid values: Type: Specify the type of scaling policy behavior. Pod: performs scaling based on the number of pods. Percentage: performs scaling based on the percentage of pods. Threshold: Specify the threshold for scaling. This parameter needs to be used together with Type. For example, if you set the Type parameter to Pod and the threshold to 10, it indicates that 10 replicas will be scaled out or scaled in within the time interval. Time interval: Specify the interval at which scaling is performed. The value is an integer. Valid values: 1 to 1800. Unit: seconds.

Effective period

Parameters	Description
Start time	Specify the time when the IHPA starts to take effect, which is accurate to seconds.
End time	Specify the end time of the IHPA. Valid values: Permanent: indicates that the IHPA never expires. Select a date: Specify the time when the IHPA ends, which is accurate to seconds. The end time needs to be not earlier than the start time.
CronHPA policy	Configure the CronHPA policy. Valid values: Repeat regularly: You can select By hour, By day, By week, or Monthly to perform scheduled scaling at a specific time point. You can also enter a Cron statement to specify the time. For configuration methods, see Cron statements. Minimal instances: Specify the minimum number of instances during scheduled scaling. Maximum instances: Specify the maximum number of instances during scheduled scaling.

Click Confirm to complete the configuration.

Result validation

Viewing overview

In the left navigation bar on the cluster management page, click Autoscaling > IHPA. On the IHPA page, click the name of the policy to go to the details page of the policy.

Click the Overview tab to view the configuration details of the intelligent scaling policy.

Viewing events

In the left navigation bar on the cluster management page, click Autoscaling > IHPA. On the IHPA page, click the name of the policy to go to the details page of the policy.

Click the Event tab to view the event details of the intelligent scaling policy.

Last updated: 2026.07.24 14:51:37