GitHub rebuilds Git storage as traffic and commits surge
GitHub is rebuilding its Git storage architecture to cope with a sharp rise in activity driven by AI coding agents. The change aims to improve write performance and reduce delays without altering developers’ workflows, review processes or security controls.
From September 2025 to August 2026, monthly GitHub traffic rose from 218.2 billion events to 473.3 billion, while September commits reached 7.38 billion, five times the year-earlier figure. The existing Spokes system waits for multiple repository replicas to confirm each write; the replacement will write to Azure Blob Storage and move read handling and maintenance work onto separate paths. Internal tests showed a 35-fold improvement in writes, but GitHub has not given a migration timeline.
- AI agent use has doubled GitHub’s monthly traffic.
- September commits reached 7.38 billion.
- GitHub’s new storage design showed 35-fold faster writes in tests.
New here? Start with this
GitHub is where software developers store and manage code for their projects. It uses technology called Git to organise and store millions of code repositories. In recent months, the amount of work flowing through GitHub has increased significantly.
The surge is being driven by artificial intelligence tools that automatically generate and suggest code. More developers are using these AI systems to help create code more quickly and efficiently. As a result, the rate at which code is being generated and stored on GitHub has accelerated dramatically.
GitHub is now rebuilding how it stores and retrieves code to cope with this increased workload. The new system will handle code submissions far more quickly and efficiently. The changes are designed to keep the platform working smoothly without affecting how developers use it or its security features.
Read the full article at the source →
Originally published by The Register as “So long, Spokes: GitHub rewrites storage to restore reliability, just in time for agentic hordes”.