Create and manage Instance groups

Reviewed on August 12, 2026

An Instance group is a collection of identical Instances, created from an Instance template and managed together. A group can run at a fixed size for predictable workloads, or scale automatically between a minimum and maximum number of Instances based on usage metrics.

When you attach a Load Balancer to an Instance group, Scaleway keeps its backend up to date as the group scales, and can automatically replace Instances that fail their health check.

Before you start

To complete the actions presented below, you must have:

• A Scaleway account logged in to the console
• Owner status or IAM permissions allowing you to perform actions in the intended Organization
• An Instance template with a CPU-only configuration

Understanding Instance group permissions

Instance group actions rely on the same permission sets as standalone Instances: InstancesReadOnly and InstancesFullAccess.

Attaching or editing a Load Balancer or Private Network for an Instance group also requires the corresponding LoadBalancersFullAccess and/or PrivateNetworksFullAccess permission sets.

With InstancesReadOnly, you can:

• View the list of Instance groups, and open the Overview, Instances, Logs, and Settings tabs of a group.
• Filter and sort the Instance groups list.
• Copy IDs and other information displayed on the pages of a group.
• Close banners and alerts.
• Switch between Availability Zones or Regions.

With InstancesReadOnly alone, you cannot:

• Open the Instance group creation wizard or the Edit settings form.
• Create an Instance group or an Instance template.
• Edit an Instance group settings, or delete an Instance group or one of its Instances.

To perform these actions, you need the InstancesFullAccess permission set, or broader access such as Owner status.

Create an Instance group

You can start the Instance group creation wizard from 3 places in the console: the Groups tab, the Templates tab, or the Overview page of an Instance template.

1. In the Compute section of the side menu, click CPU & GPU Instances. The Instances dashboard opens.

2. Open the creation wizard using one of the following methods:

   • Click the Groups tab, then click Create Instance group.
   • Click the Templates tab, click more icon next to the template you want to use, then click Create Instance group.
   • Click the Templates tab, click the name of the template you want to use, then in the Instance groups section of its Overview page, click Create Instance group.
   Tip

   You can also access this wizard right after creating a new Instance template: in the dialog box, click Create Instance group instead of Skip for now.

3. Enter a name for your Instance group, or keep the randomly-generated name. Optionally, add tags.

4. Select the Availability Zone and the Instance template to use. Every Instance in the group is created from this template.
   The console supports CPU-only Instance templates for Instance groups. GPU templates are available via the API or CLI, in fixed-size or autoscaling mode. However, autoscaling is not yet effective for GPU-based groups, since the currently supported metrics — CPU and RAM usage — are not meaningful for GPU workloads.

5. Under Choose capacity management, select whether the group runs at a fixed size or uses autoscaling, and configure the corresponding settings. See Choose capacity management for details.

6. (Optional) Enable Use Load Balancer for traffic routing to register the Instances of the group as backend servers and enable autohealing. See Configure a Load Balancer for traffic routing and autohealing for details.

7. Check the Environment footprint details.

8. In Summary, review the estimated cost of your Instance group.

9. Read and accept the Scaleway terms and conditions, then click Create Instance group.

Your Instance group is created, and you are taken to its Overview tab, where its Instances begin provisioning.

Choose capacity management

An Instance group runs in either fixed size or autoscaling capacity mode. Both modes rely on horizontal scaling (adjusting how many Instances are running) rather than vertical scaling (resizing the resources of an existing Instance).

You choose the mode when you create the group, and can switch between modes later from the group Settings tab.

Fixed size

A fixed-size group always runs the same number of Instances, making it best suited to predictable workloads. Since it never automatically reduces its Instance count, you must stop its Instances manually to avoid billing costs during periods when you do not need the fleet.

• Number of Instances: set how many Instances the group runs. You can set between 1 and 50 Instances.

Autoscaling

An autoscaling group dynamically adjusts its Instance count between a minimum and maximum value, based on the usage thresholds you define. This mode is best suited to unpredictable workloads.

• Minimum Instance count and Maximum Instance count: set the scaling range for the group (between 1 and 50 Instances).
  
  

• Scaling targets: for each metric, set a Target usage percentage. This target represents the usage level the group aims to maintain, not a strict on/off threshold. The group uses this target to calculate how many Instances it needs. A scaling event only happens when this calculation changes the number of Instances.
  

  For this reason, usage can stay above or below the target without triggering any change. For example, a group can stay stable at 82% usage with an 80% target, or at 70% with the same target, if the number of Instances stays the same, or if a cooldown period from a previous scaling event is still active.
  Click Add metric to monitor an additional metric, or click delete icon next to a metric to remove it.

  Note

  CPU usage and RAM usage are the only metrics currently supported in the console. When you set several metrics, the group scales to the maximum Instance count required by any of them.

• Scaling behavior: configure how the group reacts for each scaling event.

  • Scale out: set how many Instances to add when scaling out.
  • Scale in: set how many Instances to remove when scaling in.
  • Cooldown period: set how long the group waits after a scaling event before it can scale again. This gives new Instances time to start and stabilize before further scaling changes occur, so set it long enough to ensure they are ready to serve traffic.

Configure a Load Balancer for traffic routing and autohealing

Attaching a Load Balancer to an Instance group lets Scaleway automatically register the Instances of the group as backend servers, and route traffic to them as the group scales.

1. In the Instance group creation wizard, or from the Settings tab of the group, enable Use Load Balancer for traffic routing.
2. Select an existing Load Balancer.
   Only Load Balancers in the same Availability Zone as your Instance group are available for selection. Make sure the Load Balancer you select has enough capacity and quota for the Instances of your group.
   Creating a Load Balancer from the Instance group flow is not supported. Create a Load Balancer beforehand if you do not have one yet.
3. For each backend, select the Network and IP type to use to route traffic to your Instances.
4. (Optional) Enable Autohealing. If a backend server fails its health check, the Instance group automatically replaces it.
   You cannot enable autohealing on an Instance group that does not have a Load Balancer configured for traffic routing.

While a Load Balancer or one of its backends is in use by an Instance group, the Delete action is disabled on the Load Balancer pages. Attach a different Load Balancer to the group before you can delete the one currently in use.

Manage an Instance group

Each Instance group has the following tabs: Overview, Instances, Logs, and Settings.

View Instance group information

The Overview tab of an Instance group shows its status (Stable, Scaling up, Scaling down, or an error status), its Instance group ID, creation date, and Availability Zone, along with the following:

• Instance template used to create the group, and its ID.
• Capacity management mode, group capacity, scaling thresholds, and cooldown period.
• Public Network and Private Network information, inherited from the Instance template.
• Placement group, if the Instance template includes one.
• Tags you assigned to the group.

The placement group and tags inherited from the Instance template cannot be changed from an Instance group. Create a new Instance template with the settings you need, and use it to create a new Instance group.

View and manage the Instances in a group

1. Click the Instances tab of your Instance group. For an autoscaling group, an Autoscaling status card displays the current, minimum, and maximum Instance count, as well as the scaling status of the group.

2. Review the list of Instances belonging to the group.

3. Click more icon next to an Instance:

   • Click More info to open the Instance details.
   • Click Reboot to restart the Instance. Confirm by clicking Reboot Instance in the dialog box that appears.
   • Click Delete to remove the Instance from the group. Optionally, select Delete associated IPs and/or Delete Block Storage volumes, type DELETE to confirm, then click Delete Instance.
     
   Important

   Deleting an Instance is irreversible. The Instance group always provisions a new Instance if the count drops below its configured minimum, regardless of whether autohealing is enabled. Autohealing covers a different case — it automatically replaces an Instance that fails the Load Balancer's health check.

View Instance group logs

The Logs tab of an Instance group gives you access to Scaleway Cockpit, which ingests the metrics and logs of the group.

• Click Open Grafana logs dashboard to view and monitor your Instance group logs in a Grafana-managed dashboard.
• Click Manage custom data with tokens to create a token and send data from sources other than Scaleway to the same Cockpit.

Edit Instance group settings

1. Click the Settings tab of your Instance group.
2. Review the current Capacity management and Traffic routing & autohealing configuration.
3. Click Edit settings.
4. Update the minimum and maximum Instance count, scaling thresholds, cooldown period, and/or Load Balancer and autohealing configuration as needed.
5. Click Save settings to confirm your changes.

Delete a single Instance group

1. Click the Groups tab of the Instances dashboard.
2. Click more icon next to the Instance group you want to delete, then click Delete. Alternatively, from the Overview or Settings tab of the group, click Actions > Delete.
3. Type DELETE to confirm, then click Delete Instance group.

Delete several Instance groups at once

1. Click the Groups tab of the Instances dashboard.
2. Select the Instance groups you want to delete.
3. In the selection toolbar that appears, click delete icon.
4. Type DELETE to confirm, then click Delete Instance groups.