ClusterD provides a shared resource layer for distributed applications. Its two-level model keeps resource coordination in the cluster while frameworks make workload-specific placement and lifecycle decisions. That separation makes room for container services, batch and HPC jobs, GPU workloads, and specialised executors in one heterogeneous cluster.
Microservice environments combine services with different runtime, scaling, placement, and operational requirements. ClusterD lets them share CPUs, memory, ports, and other resources without forcing every workload through a single application scheduler. Framework schedulers can apply the logic that fits their own services while ClusterD coordinates the underlying node pool.
Run distributed applications on a dynamically shared pool of heterogeneous nodes. CPU, memory, GPU, storage, network and custom resources can be represented and governed with roles, weights, reservations and quotas.
ClusterD managers expose available resources to registered frameworks. Framework schedulers answer the workload questions—where, when, how many, and which resources—while executors perform the work.
Containers are one workload type, not the platform boundary. Use the native Mesos® containerizer or Docker containerizer, and build other execution paths through framework and executor APIs.
High-availability masters, maintenance primitives, quotas, reservations and observability provide the operational foundation for a shared cluster resource pool.
SSL support, authentication, authorization, and secrets management are available platform capabilities for deployments that need controlled access to cluster infrastructure.
ClusterD continues Apache Mesos® as an open, framework-driven cluster manager. Teams can inspect the implementation and adapt the scheduler layer to their own needs.
ClusterD can serve as the shared infrastructure below different application platforms. OCI, CNI and CSI keep container, network and storage integration open, while framework APIs keep workload policy extensible.
mesos-compose translates its documented Compose 3.9 feature set into ClusterD workloads, including replicas, resource limits, ports, volumes, networks, health checks, and placement constraints.
mesos-m3s runs K3s/Kubernetes components as framework-managed workloads and supports multiple independent framework instances on ClusterD.
Scheduler and executor APIs allow teams to build or extend a framework when their services need their own placement and lifecycle logic.
