Selecting Instance Types¶

Note

This feature is currently experimental, with new features under active development.

When creating clusters, Coiled will match the instance type with the requirements that you specify for CPU, Memory and GPU. Coiled chooses five instance types to serve as a fallback when creating a Cluster since sometimes a specific instance type might not be available in your cloud provider of choice.

You might not wish to get allocated random instance types, and you might want to provide a list of instance types when creating a Cluster. This will allow you to have more fine-grain control of the type of Cluster that you create.

With the keyword argument scheduler_vm_types and worker_vm_types, you can specify instance types for both the Scheduler and Workers. For example:

import coiled

cluster = coiled.Cluster(
    scheduler_vm_types=["t3.large", "t3.xlarge"],
    worker_vm_types=["m5n.large", "m5zn.large"],
)

It’s recommended that you specify more than one instance type in your list to avoid instance availability issues in the cloud provider and region that you are using Coiled.

Note

The order of the instance type will not be preserved when creating the cluster.

Instance types allowed¶

Currently, you are allowed to choose an instance type from a subset of all the instance types available in your cloud provider of choice. In this list, you have the most common instances used by users when creating Clusters with Coiled.

You can use the command coiled.list_instance_types() to see a list of all allowed instance types across all cloud providers that you can use.

import coiled

coiled.list_instance_types()

Alternatively, you can specify which cloud provider you wish to use.

import coiled

coiled.list_instance_types("aws")

You might be interested in reading Selecting GPU Types.

Setting Backend Options via Python API

Selecting GPU Types