LinkedIn, the professional networking site, built its own deployment engine to replace Argo CD, an open-source GitOps tool that keeps Kubernetes clusters in sync with configuration stored in Git. LinkedIn runs more than 250 clusters, 4,000 services and 100,000 deployments a week. With one Argo CD instance per cluster, sync slowed as pod counts grew, and some apps showed "Synced and Healthy" at the wrong revision. The new engine runs each deployment in its own pod, with no shared controller state. The migration ran from July 2025 to April 2026.
The company says infrastructure failures fell from 925 to 63 while deployment volume rose 3.7x. Median time to detect a failed deployment dropped from 16.1 to 2.45 minutes. LinkedIn says orchestrator improvements also contributed. It rejected Flux, a similar tool, because it shares the same design. The engine is not open source, and Argo CD still runs cluster add-ons.
