GitHub’s scaling problem is handling huge numbers of relatively small repos at high concurrency. This is not the same problem as Google, Microsoft, Meta et al w

GitHub’s scaling problem is handling huge numbers of relatively small repos at high concurrency. This is not the same problem as Google, Microsoft, Meta et al who have terabyte-scale monorepos, and all have lazy-load solutions to create virtual clones where only the parts you are working on are loaded onto local disk. Microsoft Windows moved to GitHub some time ago and they used the filesystem approach OP dismisses too blithely: https://github.com/microsoft/scalar https://github.com/microsoft/scalar