The Technical Oversight Committee (TOC) of the Cloud Native Computing Foundation (CNCF) has voted in favor of accepting the Open Cluster Management (OCM) project as an Incubating Project. This step reflects the architectural maturity achieved by the project and the expansion of its adoption in production environments for managing distributed Kubernetes clusters across the cloud, local data centers, and edge computing environments.
An Architectural Structure Based on Independence and Security
The OCM project relies on a hub-and-spoke architecture using declarative Kubernetes APIs without imposing a proprietary control platform. This architecture allows for cluster registration, workload distribution, and software policies with high flexibility, while ensuring the continued operation of subordinate clusters even in the event of a lost connection to the main control center or the failure of one of its components.
The ecosystem is characterized by the decentralization of sensitive credentials; the project relies on an asynchronous pull model, meaning that the operating agent installed on the managed cluster fetches tasks from the hub rather than the hub having direct access to each cluster, which raises security readiness and prevents permission leaks.
Ecosystem Components and Development Path
The project originally started through a collaboration between IBM and Red Hat, and joined the foundation's sandbox stage in late 2021. Following the release of major version 1.0 in June 2025, the platform saw four minor updates up to version 1.4 in October 2026, including integrations with gRPC protocols, OpenTelemetry libraries, and continuous deployment tools.
The platform consists of core modules that manage distribution tasks:
- Cluster Manager: The central control panel on the hub for coordinating registration and resource distribution.
- Klusterlet Agent: The software client installed on each managed cluster to execute instructions.
- ManifestWork API: To distribute Kubernetes resources and handle conflicts via the pull mechanism.
- Placement API: To dynamically route workloads based on available capacity and labels.
- Add-on framework: A software environment allowing other cloud tools to easily integrate with multi-cluster environments.
What This Development Means for Infrastructure Teams
The project solves the problem of operational complexity as enterprise infrastructures grow and distribute geographically; global platforms like eBay rely on OCM to deploy applications across more than 190 Kubernetes clusters in cooperation with the Argo CD tool. Artificial intelligence platforms, such as VESSEL AI, also employ it to manage GPU clusters and distribute model training and execution workloads across multiple clouds, providing an open-source standard alternative that prevents organizations from vendor lock-in with closed cloud service provider solutions.