HeadlinesBriefing favicon HeadlinesBriefing.com

Docker Pulls Waste Bandwidth - Tool Shows How to Fix It

Hacker News •
×

A developer has created a tool to visualize how inefficient docker pull operations can be, particularly problematic for robotics deployments and other scenarios with many dependencies. The tool demonstrates that updating a single dependency often drags along gigabytes of unchanged files, wasting bandwidth and time. This inefficiency becomes critical when pulling updates to devices in remote locations with poor internet connections.

Docker's current layer-based architecture invalidates entire layers when even a single byte changes, forcing redundant downloads of byte-for-byte identical files. The developer's analysis shows that a registry capable of deduplicating at the file level rather than just the layer level could significantly reduce storage and bandwidth costs. The tool allows users to compare two Docker images and estimate potential time savings on various internet speeds.

The developer is building a new registry and transport system that will enable file sharing across Docker layers both locally and in the registry. While some prior art exists with the eStargz format, which splits filesystem metadata from contents while remaining OCI compliant, it focuses on lazy pulls without deduplication. The new system aims to compete with existing image providers on both cost and speed by addressing the fundamental inefficiency in Docker's current approach.