distributed systems

Networked servers representing the focus_keyword challenge in Git infrastructure scaling
Infrastructure

The Last 25 Milliseconds: What GitHub’s Move to Azure Really Teaches Us About Scaling Git

Picture a courier who has to deliver every letter in a city, one at a time, waiting for a signed receipt from every house before setting off with the next envelope. If one house is a thousand miles away, every single letter in that courier’s bag gets stuck behind the slowest trip, no matter how close the other houses are. That, in miniature, is the problem GitHub has been quietly solving while shifting an enormous share of its Git infrastructure onto Microsoft’s Azure cloud.

A server room and laptop showing a code repository used as a focus_keyword for AI coding agents
Tech Market

Why an Ex-GitHub CEO Thinks Code Hosting Needs to Be Rebuilt for Robots, Not Humans

When a company’s traffic pattern changes from “a person checks in occasionally” to “a fleet of software agents hammers the same repository thousands of times an hour,” the plumbing built for the first scenario starts to strain under the second. That is the bet behind Entire, a startup founded by Thomas Dohmke, the former CEO of GitHub, which this week opened a preview of a distributed network for mirroring code repositories. The pitch is blunt: even the platform Dohmke used to run cannot keep up with how AI coding agents actually use Git.

Scroll to Top