Container Image Compression via Block Device Mounting

The patent introduces a local disk as an intermediary between the container and the network file system. The image is first pulled and stored locally on the disk, then mounted to a local directory. This local disk intermediary eliminates network transmission delays for subsequent access while maintaining the ability to use network storage for the base image, thus resolving the contradiction between reducing image size and maintaining user experience.

The patent segments the image access into two parts: the base image stored in the network file system and the locally cached image layers. This segmentation allows the system to maintain a single comprehensive image definition while only transferring and storing the necessary layers locally, reducing both single image size and maintenance complexity compared to splitting into multiple functional images.

The patent implements preliminary action by pre-pulling and caching image layers to the local disk before they are needed for container execution. The first-time pull operation stores layers locally, and subsequent container creations reuse these cached layers. This preliminary caching action reduces network resource pressure during high concurrency while maintaining fast distribution speed for subsequent operations.