DKube offers a full MLOps workflow regardless of whether the job runs on the local Kubernetes cluster or on a remote Slurm cluster. MLOps on Slurm capability is provided without any compromise on the Kubeflow foundation of DKube or on the extended capabilities offered beyond the standard framework.
The remote cluster is easily added and managed from the DKube UI.
The program code and datasets do not need to be changed in order to submit them to the remote Slurm cluster. All required translation is handled by DKube, and the remote execution supports all of the powerful features of the platform, including tracking, lineage, and the metric display and compare features based on MLFlow.
Models generated from a remote Slurm execution use the same workflow as locally-created Kubernetes models.