Skip to main content
Version: Latest

Overview of Anyscale Cloud

Check your docs version

These docs are for the new Anyscale design. If you started using Anyscale before April 2024, use Version 1.0.0 of the docs. If you're transitioning to Anyscale Preview, see the guide for how to migrate.

An Anyscale Cloud abstracts the resources and infrastructure necessary for managing Anyscale clusters. It performs the following functions:

  • Maintains information on where and how to start clusters.
  • Deploys clusters in your AWS or GCP account.
  • Defines the collection of resources necessary to manage Anyscale clusters.

When you sign-up, Anyscale provisions an Anyscale-hosted cloud for you to start running workloads. The setup and maintenance of the underlying infrastructure has been provided for you.

For more control, you can deploy and manage a self-hosted cloud in your cloud environment, either on AWS or GCP.

Self-hosted cloud deployment options

Self-hosted Anyscale Clouds fit into your cloud infrastructure, whatever that might look like. There are two different options to deploy an self-hosted Anyscale Cloud. The goal is to simplify the cloud setup process to improve the deployment experience. The two options are as follows:

  1. Anyscale-managed resources: The Anyscale CLI creates the required resources and configurations for Anyscale to run in your AWS or GCP account. You can quickly deploy a new cloud with the anyscale cloud setup command which uses direct networking over the internet.

  2. Customer-defined resources: You create and define resources in your AWS or GCP account with anyscale cloud register, so that Anyscale can deploy and manage clusters. In this approach, customer-defined networking allows you to run Ray workloads in private networks without accessing them over the internet to protect sensitive workloads. For fine-grained control over your cloud, Anyscale provides a Terraform module to scope down permissions given to Anyscale.