GitHub's Data Model: How to Store Version History in a Database

How do you store the entire history of a codebase — every version of every file across thousands of commits — without the database exploding? The naive answer ("save a full copy of every file on every commit") would store the same unchanged README ten thousand times. Git's answer is one idea applied relentlessly: address every object by the hash of its own content. Identical content always lands at the same address, so it's stored exactly once — and an unchanged file is automatically shared across every commit that contains it. This is the data model behind Git, and behind GitHub at scale. The…