AdministrationPlatform

Compute

Where node runs and deployments execute, and how the ones that need a GPU get one.

Every node run and every deployment executes as a container. Where those containers start is set once per installation by an instance administrator, under Admin → Configuration → Executor. On OvalBee Cloud, compute is provided and this page is background reading.

Executor backends

BackendHow it worksGood for
DockerContainers start on the Docker host OvalBee runs on.A single powerful machine, evaluations, small teams.
KubernetesEach run is a pod in a Kubernetes cluster, placed by the cluster’s scheduler.Teams sharing a pool of GPU nodes, scaling out, separating workloads.

With Kubernetes, the cluster can be the one OvalBee itself runs in, or a separate cluster — for example a GPU cluster in another account or on premises, while OvalBee runs elsewhere. The separate cluster gets a small executor-only installation, and the main installation is given its API address and credentials. Deployments started on a separate cluster need to reach the OvalBee API over the network.

Machine classes

Nodes don’t pick hardware — they ask for a machine class. Nodes that train or run models ask for the gpu class, and everything else runs on the default CPU class. The administrator decides what each class means on the chosen backend:

  • On Docker, the gpu class runs containers with the NVIDIA runtime, so the GPUs must be on the Docker host with the NVIDIA container toolkit installed.
  • On Kubernetes, each class maps to pod settings: resource requests such as one nvidia.com/gpu, node selectors, tolerations and affinity. That’s how you steer training to your GPU nodes and keep everything else off them.

You can define more classes than gpu - for example a larger one for heavy training — in the executor options, which are edited as JSON for now. The Deploy dialog lists the installation’s classes for models that let you choose the device.

What users notice

  • Nodes that need a GPU are marked with a GPU badge in the Node Library.
  • The first run of a node on a machine can take a little longer while its container image is pulled.
  • A deployment holds its machine for as long as it runs, because the model stays loaded in memory. Stop deployments you don’t use to free the GPU.
  • Activity in the main sidebar shows everything running and waiting across the workspace.