The development

What happened

GitHub’s 21 October engineering report describes an Azure Git-storage migration for rising traffic from developers and AI coding agents. These agents implement and verify software changes, adding work to shared repositories. GitHub says a copy of 99.99999% of its repositories has moved to Azure; proposed per-reference quorums and durable Azure Blob Storage address the remaining migration and separate read capacity from writes.

Key facts

  • GitHub’s blog says a copy of 99.99999% of its repositories has been moved to Azure; it identifies the remaining 0.00001% as the migration challenge.

  • GitHub says traffic growth, data-center and regional power limits, and physical supply constraints shaped the migration’s timing and pace.

  • The post says per-reference quorums are intended to bring the remaining repositories over and enable replicas without persistent state of their own.

  • GitHub says durable storage in Azure Blob Storage could decouple reads from the write quorum, allowing read and write compute to scale independently.

  • In a later 28 October statement, GitHub explicitly connects AI assistants and agents that implement and verify changes with its continuing Git-infrastructure investment.

The context

Why it matters

AI coding workflows depend on ordinary repository infrastructure as well as models. The 21 October report describes capacity constraints and promising internal read tests, without a numerical read-throughput comparison or a timetable for its remaining repositories. The later statement supplies AI-agent context, not independent proof of those engineering results.