My application stores large binary files (100Mb - 1Gb) that change over time on s3. I want to offer the ability to roll back or download previous versions of these files without having to store snapshot copies.
What would be the best way to achieve this?
Since Nexus is able to work with s3 repositories (see ticket NEXUS-1121), I would advise you to record those different versions in an artifact repo like Nexus, and not in a VCS (Version Control System tool like SVN, or even worse, distributed ones like Git or Mercurial).
A VCS has features (branching, merging, diff) that large binaries won't take advantage of.
