Shared resource pool
Masters expose available CPU, memory, ports and other resources to registered frameworks across a dynamically shared node pool.
The project
Cluster infrastructure, continued.
ClusterD carries the Apache Mesos® architecture forward as an open resource-orchestration platform for heterogeneous workloads and distributed frameworks.
Overview
A shared resource layer for independent schedulers.
ClusterD provides efficient resource isolation and sharing across distributed applications, or frameworks. It makes a dynamically shared pool of heterogeneous nodes available to framework schedulers, while each framework retains the workload-specific logic for placement and lifecycle decisions.
The project continues Apache Mesos® as a dependable, open foundation for modern Linux and container workloads. The core stays focused on resource coordination, scheduling APIs and task execution; capabilities such as container, network and storage integration can build on OCI, CNI and CSI.
Core capabilities
One cluster, many ways to run work.
ClusterD keeps resource coordination central without prescribing a single application platform.
Two-level scheduling
ClusterD coordinates resources and offers; framework schedulers decide which workload-specific tasks should run.
Containerizer choice
Run workloads with the native Mesos® containerizer or the Docker containerizer, depending on the framework and deployment.
Framework ecosystem
Use Kubernetes through mesos-m3s, Compose-defined services, Airflow integrations, Toil workflows or a custom framework.
Development
Open infrastructure with a practical path forward.
ClusterD is developed as an ongoing continuation of Apache Mesos®. The project maintains the core framework-driven model while improving the Linux, container and operational foundations needed by current deployments.
Build instructions, architecture notes and operational references are maintained in the project documentation. The modern alternative WebUI is developed separately for ClusterD and Apache Mesos®.