GitHub details infrastructure scaling after major August outage

GitHub experienced a 7-hour 47-minute global outage on August 17 caused by scaling failures in its Central US data center during traffic peaks. The incident disrupted core services including authentication, GitHub Actions, APIs, and Copilot—following another major outage on August 6. Since April, the platform processed 2.9 billion monthly commits (up from 1.4 billion), exposing architectural bottlenecks. GitHub added 3M+ CPU cores and 120PB storage while migrating 58% of platform load to Azure. Ongoing fixes include linear read scaling for monorepos and retry storm protections.