First release candidate for Docker 1.10 appeared

Up to now, each image and each layer used a randomly assigned Universally Unique Identifier (UUID). On the other hand, Docker 1.0 relies on an associative method which uses an ID based on a secure hash of the image and layer data. The new method promises more security, should eliminate ID collisions and guarantee data integrity after pull and push, as well as load and save.

Addressing images by their content will not only make it easier for the user to find out whether something has already been downloaded, but also builds the basis of the new distribution feature. The pull and push code now follows a download or upload manager concept which offers significantly improved stability and should eliminate potential issues with parallel requests. The possibility to retry failed downloads and improved prioritization of concurrent downloads has also been offered.

Read also: 5 steps to becoming a quality Docker contributor

Last but not least, Docker 1.10 carries a new manifest format with it. It directly references the content addressable image configuration, as well as the layer checksums and makes it possible to use a manifest list for multiple architectures or platforms.

You can find further information on how to prepare your images for migration to the new model in the corresponding press release. In addition, a migration tool is available on GitHub which can even be executed with a running daemon, thus reducing necessary downtime.