Automotive Linux Summit & Open Source Summit Japan 2018

The Git version control system was created to manage the development of the Linux kernel - a large software project in many respects. But it struggles to scale beyond that, to Enterprise scale “monorepos” or with repositories with large files like binaries, media, or scientific data sets.

This session will introduce some of the architectural decisions in Git that lead to its struggle with large files or very large repositories, and show techniques to cope with them. Edward will discuss:

* Why Git (by itself) is not appropriate as a storage mechanism for large files, and how tools like git-lfs work and can allow you to manage binaries.
* Why Git cannot handle enormous “monorepos”, and how tools like GVFS work to enable them.

This session will provide evidence of teams using these tools successfully and show examples of how they work.