Accelerate AI container image pulls by using a VPC
The ACS container image service is accessible over the public network. When you start a workload on an ACS product, pulling the image over the public network can be time-consuming. To improve performance, ACS products provide a VPC-based image access service. Use this service to accelerate your AI container image pulls.
Applicability
You can specify the VPC URI for an AI container image when you create a workload in the ACS console, or specify it in a YAML file.
In non-ACS environments, the VPC method is unavailable.
You do not need to authenticate to pull PPU AI container images in this scenario. You also do not need to add a Secret for authentication or specify the
imagePullSecretsfield when you create a Pod.Specifying the
imagePullSecretsfield has no effect. You do not need to modify your existing YAML files.
How to use
Scope
All existing and new PPU AI container images in theegslingjun-registry.cn-wulanchabu.cr.aliyuncs.com/egslingjun repository support IN-VPC image pulls.
Method
Replace the image URIegslingjun-registry.cn-wulanchabu.cr.aliyuncs.com/egslingjun/{image:tag} withacs-registry-vpc.{region-id}.cr.aliyuncs.com/egslingjun/{image:tag} to pull PPU AI container images from within your VPC.
{region-id}: The ID of the service region for your ACS product. This includes financial cloud and government cloud regions. For example:cn-beijing,cn-wulanchabu, andcn-shanghai-finance-1.{image:tag}: The name and tag of the AI container image. For example:inference-xpu-pytorch:25.11-v1.7.0-vllm0.10.2-torch2.8-cu129-20251113andtraining-xpu-pytorch:25.11.
Example
Original image URI for pulls over the public network:
egslingjun-registry.cn-wulanchabu.cr.aliyuncs.com/egslingjun/training-xpu-pytorch:25.04New image URI for accelerated IN-VPC pulls:
# For an ACS product in the China (Beijing) region
acs-registry-vpc.cn-beijing.cr.aliyuncs.com/egslingjun/training-xpu-pytorch:25.04
# For an ACS product in the China (Ulanqab) region
acs-registry-vpc.cn-wulanchabu.cr.aliyuncs.com/egslingjun/training-xpu-pytorch:25.04You can still use theegslingjun-registry.cn-wulanchabu.cr.aliyuncs.com/egslingjun/{image:tag} image URI to pull images over the public network.