Git at Scale

The Git version control system was purpose built to handle a single source code base: the Linux kernel. As a result, many of the assumptions and architectural decisions that Git has made have been focused on repositories that look like the Linux kernel source tree. As a result, Git has problems at enterprise scale, and struggles to operate on source trees that have very large files, have many people working in it, or have an immense number of files.

Edward Thomson will break down solutions to each of these problems, based on his experience as a software engineer working on GitHub and Visual Studio Team Services. He'll introduce techniques to include large binaries in your repository, at ways that Git hosting providers scale to handle many simultaneous users, and scaling Git to handle the largest codebases in the world.