Download
The DaoCloud Public Image Mirror project provides a stable and efficient container image acceleration service designed to improve download speeds in regions where access to global registries is slow, such as China. Many popular registries like Docker Hub, GCR, and GHCR are located overseas, often causing significant delays when pulling images. This project solves the problem by offering mirrored registries with consistent image hashes and real-time synchronization. It uses a lazy-loading mechanism, caching image layers in third-party object storage, ensuring that frequently used images are delivered faster. The system is simple to use, requiring only a prefix replacement to pull images from the mirror instead of the original registry. By reducing latency and improving reliability, the service supports developers in accelerating Kubernetes, Docker, Containerd, and AI model image downloads in production environments.
Features
- Provides mirror acceleration for major registries like Docker Hub, GCR, and GHCR
- Ensures all image hashes remain consistent with the source registry
- Uses lazy-loading and caching with third-party object storage for efficiency
- Offers simple prefix replacement for seamless integration
- Supports Kubernetes, Docker, Containerd, and Ollama acceleration setups
- Delivers stable and real-time synchronization with minimal delays