Resource pool

更新时间:
复制 MD 格式

Use the resource pool to purchase and manage AI computing resources in PAI. You can create resource groups, purchase cloud-native resources, and manage renewals and unsubscriptions in one place.

Key concepts

Create a resource group and purchase resources

The following cloud-native resource types are available when you create a resource group:

Resource type

Description

Included resources

Create a resource group and purchase Lingjun resources

High-performance computing resources for large-scale AI training and inference. Built on co-optimized hardware and software, with RDMA high-speed networking and storage acceleration. Available to allowlisted users only. Contact your account manager to request access.

Compute nodes

Create a resource group and purchase general computing resources

Flexible and stable computing for small-to-mid-scale AI training and inference, powered by ECS and other cloud products. Available immediately after activating PAI.

Compute nodes , cloud disks (data disks)

Note

General computing resources support purchasing compute nodes and cloud disks (data disks) together:

  • Compute nodes: Required. Select the resource specifications for your compute nodes.

  • Cloud disks (data disks): Optional. Provide additional storage for compute nodes. For details, see ESSDs.

Manage resources

Log in to the PAI console and go to the Resource pool page. Click a resource group name to view and manage its purchased resources. Node orders are split by instance so you can manage each node individually — for example, to renew or unsubscribe.

Add tags to node orders to track costs by tag. To view billing details by tag, see Use tags for cost allocation.

Billing

AI computing resources use a subscription (prepaid) billing model. Charges are based on the node specifications, quantity, and subscription duration. For details, see Billing of AI computing resources.

What's next

After creating a resource group and purchasing resources, allocate the resources to a resource quota for AI development or training. For details, see Overview.