This page describes the top-level Anyscale concepts that you see when logging into the console.
A defined deployment area in your AWS accounts or GCP projects where you can launch Ray clusters.
Anyscale supports multi-cloud setup and your organization can choose to deploy multiple Clouds in one cloud provider or across different cloud providers. View architecture for more details about how Anyscale Clouds are deployed.
An Anyscale Account represents an organization where administrators can manage users, credentials, billing, and other top level configurations.
Workspaces are fully managed development environments that let you program the cluster by using familiar tools like Visual Studio Code or JupyterLab Notebooks. They give you auto-scaling compute resources, friendly package distribution, and seamless code-test-debug experience in the Anyscale cloud.
Discrete batch operations managed by Anyscale and running in their own Ray clusters with full cluster lifecycle management, retry capabilities, alerting, and scheduling.
Anyscale Services serve your models behind endpoints in a high-available manner with support for zero downtime upgrade, performance monitoring, alerting, etc.
Anyscale Clusters are managed Ray clusters started by Anyscale for your Workspaces, Jobs, and Services. You can also manually create and interact directly with them.
Compute Configs specify the hardware settings of your cluster, including but not limited to Anyscale Cloud, node types, autoscaling configurations, etc.
Cluster Environments specify the software dependencies of your cluster, including but not limited to base images, environment variables,
pip dependencies, etc. They can be built on top of the Anyscale-supplied base images or your own docker.
If a Project is not specified, resources are created without a Project and accessible to all the users of that Cloud.
Throughout this documentation, you'll see references to the four interfaces that Anyscale supports: