Overview of EAS resources - Platform For AI - Alibaba Cloud Documentation Center

Elastic Algorithm Service (EAS) allows you to use public resources, EAS resource groups, and resource quotas. EAS resource groups are divided into dedicated resource groups and virtual resource groups.

Comparison between resource types

Resource type		Scenario	Billing method	Feature
Public resources		Public resources are suitable for test scenarios or businesses with obvious traffic fluctuations (dedicated resources and elastic resource pools).	You are charged after you use the resources. You can purchase public resources based on your business requirements. Pay-as-you-go. For more information, see Billing of EAS.	Provides computing resources that are available without purchase. Resource supply is not guaranteed during peak hours. Provides CPU and GPU resources. The GPU resource types include A10, P4, P100, T4, and V100.
EAS resource groups	Dedicated resource groups	Dedicated resource groups are suitable for scenarios that require resource isolation or have high security requirements. For resources that are in short supply, you can purchase them as dedicated resources to ensure resource availability.	You must first purchase the resources before you can use them. Subscription and pay-as-you-go. For more information, see Billing of EAS.	Provides isolated computing resources that are exclusive to you, which ensures data security. Provides CPU and GPU resources. The GPU resource types include A10, P4, P100, T4, and V100. Provides the GPU sharing feature (available only for users in the whitelist).
EAS resource groups	Virtual resource groups	Virtual resource groups are suitable for scenarios that require different types of resources, including public resources, resource quotas, and EAS dedicated resources.	You are charged based on the resources that are actually used during service scheduling.	Supports service deployment by using multiple types of resources. Supports the setting of the priority for different resource types.
Resource quotas		Only Lingjun resource quotas can be used.	You must first purchase the resources before you can use them. Subscription. For more information, see Billing of AI computing resources.	Manages integrated resources for model training and inference and implements resource utilization.

References

The EAS module of Platform for AI (PAI) provides an end-to-end platform for model development and deployment. For more information, see EAS overview.
For information about the scenarios, billing, and usage of the public resource group, see Use public resources.
For information about the types, operations, and management of dedicated resource groups, see Work with EAS resource groups.
You can collect logs of an EAS service and store them in your Simple Log Service logstore. For more information, see Configure log collection for a resource group.
After you configure a virtual private cloud (VPC) for a resource group, clients can access online EAS services that are deployed in the resource group, and the online EAS services can access other Alibaba Cloud services that reside in the same VPC. For more information about network connectivity configuration, see Configure network connectivity.