Using Instance Configurations and Instance Pools

An instance configuration defines the settings to use when creating compute instances, including details such as the base image, shape, and metadata. You can also specify the associated resources for the instance, such as block volume attachments and network configuration, and you can associate the instance with a capacity reservation.

For steps to create an instance configuration, see Creating an Instance Configuration.

To modify an existing instance configuration, create a new instance configuration with the desired settings.

For steps to delete an instance configuration, see Deleting Instance Configurations.

Instance pools let you create and manage multiple compute instances within the same region as a group. They also enable integration with other services, such as the Load Balancer service and IAM service.

You create an instance pool using an existing instance configuration.

After you have created an instance pool, you can update the size of the pool, add and remove existing instances from the pool, and attach or detach load balancers and network load balancers. You can also update the instance pool to use a different instance configuration, or to place instances in a different availability domain, fault domain, or subnet.

You can automatically adjust the number of instances in an instance pool based on performance metrics or a schedule. You can also stop and start instances in an instance pool based on a schedule. To do this, enable autoscaling for the instance pool.

A cluster network is a special kind of instance pool that is designed for massive, high-performance computing jobs.

By default, the instances in a pool are distributed across all fault domains in a best-effort manner based on capacity. If capacity isn't available in one fault domain, the instances are placed in other fault domains to allow the instance pool to launch successfully.

In a high availability scenario, you can require that the instances in a pool are evenly distributed across each of the fault domains that you specify. When sufficient capacity isn't available in one of the fault domains, the instance pool doesn't launch or scale successfully, and a work request for the instance pool returns an "out of capacity" error. To fix the capacity error, either wait for capacity to become available, or update the placement configuration (the availability domain and fault domain) for the instance pool.

When working with instance configurations and instance pools, keep the following things in mind:

• You cannot delete an instance configuration if it is associated with an instance pool.
• You can use the same instance configuration for multiple instance pools. However, an instance pool can have only one instance configuration associated with it.

• If an instance pool has been in the scaling or provisioning state for an extended period, the number of instances requested might exceed the number of instances available. In this scenario, after 24 hours, the instance pool transitions to Running with the available capacity, which might be less than the number of instances requested. The instance pool size updates to reflect the number of instances currently running.

  • When this happens, one of the following errors might appear in the work request: ServiceLimit, OutOfCapacity, PostLaunchFailure, or InvalidParameter.
  • If the number of instances exceeds your tenancy's service limits for that availability domain, you can check your service limits and request a service limit increase.
• If you modify the instance configuration for an instance pool, existing instances that are part of that pool don't change. Any new instances that are created after you modify the instance configuration use the new instance configuration. New instances are not created unless you increase the size of the instance pool or terminate existing instances.
• If you decrease the size of an instance pool, to balance the instances across placements (availability domain and fault domain), instances are terminated first based on how many instances from the instance pool are in that availability domain and fault domain. Within a placement, the oldest instances are terminated first.

For administrators: For a typical policy that gives access to instance pools and instance configurations, see Let users manage Compute instance configurations, instance pools, and cluster networks.

Apply tags to resources to help organize them according to business needs. Apply tags at the time you create a resource, or update the resource later with the wanted tags. For general information about applying tags, see Resource Tags.