Docker Hub outage prompted ACR failover
Diving deeper into
Joe Zeng, software engineer at Statsig, on using Docker
we did get affected by a Dockerhub outage where we couldn't pull images and so we had to spin up some images on ACR
Analyzed 3 sources
Reviewing context
This shows Docker is sticky at the workflow layer, but Docker Hub itself can still be a single point of failure. For Statsig, the outage did not force a rethink of containers or Docker Desktop. It forced a quick registry failover to Azure Container Registry, because the practical dependency was not how images were built, but where production systems pulled them from inside an Azure and AKS based deployment setup.
-
In practice, Docker Hub is the warehouse for prebuilt container images. If that warehouse is unavailable, Kubernetes clusters cannot pull the app packages they need to start new pods. Moving images to ACR is operationally much lighter than replacing Docker across developer machines and CI pipelines.
-
Statsig said switching registries would mostly mean changing URLs and API keys, and that Docker spend was still negligible. That explains why an outage triggered a backup move, not a full migration. The pain was concentrated in image distribution, while the broader Docker workflow remained good enough.
-
This fits Docker’s business model after its pivot. Docker monetized Docker Hub and Docker Desktop as paid seats for businesses, which helped revenue scale from about $11M ARR in 2020 to about $135M in 2022. That also raised the importance of Hub reliability, because the paid product sits on a daily build and pull path.
Going forward, more teams will keep Docker in the developer workflow while adding a second registry inside their cloud provider as insurance. That pushes the market toward a split model, with Docker owning the desktop and developer entry point, and cloud registries like ACR owning more of the production image distribution path.