All Products
Search
Document Center

Container Compute Service:Use VPC to accelerate the pulling of AI container images

Last Updated:Mar 25, 2026

By default, ACS pulls AI container images over the public network, which can be slow for large AI workloads. Switching to IN-VPC image pulling routes traffic through the internal network, reducing pull times significantly.

Switching to IN-VPC does not invalidate the original public network URI. The original image URI can still be pulled over the public network, so existing image references continue to work.

Supported environments

IN-VPC image pulling is only supported in ACS environments. Non-ACS environments cannot use IN-VPC access.

The following ACS pod creation methods support IN-VPC pulling:

  • Selecting images from the artifact center on the workload page in the ACS console

  • Referencing images in a YAML file

Inapplicable scenarios

Pulling AI container images through the public network is an inapplicable scenario for IN-VPC acceleration.

Pull images over VPC

All existing and new ACS AI container images in the egslingjun-registry.cn-wulanchabu.cr.aliyuncs.com/egslingjun repository support IN-VPC pulling.

Replace the public network registry host in your image URI with the region-specific VPC endpoint.

URI componentPublic networkIN-VPC
Registry hostegslingjun-registry.cn-wulanchabu.cr.aliyuncs.comacs-registry-vpc.{region-id}.cr.aliyuncs.com
Repositoryegslingjunegslingjun (unchanged)
Image and tag{image:tag}{image:tag} (unchanged)

Replace {region-id} with the ID of the region where your ACS service runs. For example:

RegionRegion ID
China (Beijing)cn-beijing
China (Ulanqab)cn-wulanchabu

For the full list of supported regions, see Regions.

Example {image:tag} values:

  • inference-nv-pytorch:25.10-vllm0.11.0-pytorch2.8-cu128-20251028-serverless

  • training-nv-pytorch:25.10-serverless

Example

The following example shows how to update an image URI for a training workload.

Original URI (public network):

egslingjun-registry.cn-wulanchabu.cr.aliyuncs.com/egslingjun/training-nv-pytorch:25.10-serverless

Updated URIs (IN-VPC), by region:

# China (Beijing)
acs-registry-vpc.cn-beijing.cr.aliyuncs.com/egslingjun/training-nv-pytorch:25.10-serverless

# China (Ulanqab)
acs-registry-vpc.cn-wulanchabu.cr.aliyuncs.com/egslingjun/training-nv-pytorch:25.10-serverless