AdministrationPlatform

Self-hosted installation

What an OvalBee installation on your own infrastructure consists of, and how to prepare for it.

OvalBee can run entirely inside your infrastructure: your cloud account, a dedicated VPC or your own data center. Images, annotations, models and every node run stay within it. During the private beta, the OvalBee team installs it together with you — this page describes what’s involved so you can have the pieces ready.

Deployment options

OptionBest for
Kubernetes with the OvalBee Helm chartProduction and teams. Scales out, and places GPU work on GPU nodes.
Docker Compose on one Linux machineEvaluations, pilots and small teams with one GPU server.

With Kubernetes, node runs can also go to a separate GPU cluster through an executor-only installation of the same chart, while the platform runs elsewhere. See Compute.

What gets installed

ComponentRole
Web app and APIThe interface and the platform itself. The API also starts node runs and deployments on the executor.
PostgreSQLFlows, nodes, runs, datasets, models and accounts.
RedisQueues, schedules and live updates.
Object storageImages, annotations, models and reports. Bundled S3-compatible storage, or your own S3, Azure Blob Storage or filesystem.
QdrantThe embedding index used by the similarity and data-curation nodes.
Labeling toolThe annotation tool labelers open from a flow.
Image proxy and CV utilitiesFast previews and thumbnails, and image processing for the interface.

PostgreSQL, Redis and object storage can each run inside the installation or be services you already operate, such as a managed database or an existing bucket.

What to prepare

  • A place to run it: a Kubernetes cluster, or a Linux host with Docker for the Compose option.
  • GPU capacity for training and inference: GPU nodes in the cluster with the NVIDIA device plugin, or a GPU host with the NVIDIA container toolkit.
  • A DNS name and TLS certificate for the address people will use. On Kubernetes you route it with your own ingress controller — the chart doesn’t impose one.
  • Storage: an S3-compatible bucket or Azure container, unless you use the bundled storage.
  • Optionally, managed PostgreSQL and Redis, your identity provider’s OIDC details for single sign-on, and SMTP for email.
  • Network access from the cluster to the OvalBee container registry, or a mirror of the images in your own registry.

Configuration

A handful of settings are needed at startup — the application secret, database, Redis and storage connections — and are passed as environment variables or Helm values. Everything else is a normal platform setting. Those can be edited in the UI, or supplied in a configuration file at deployment time so they live in version control next to the rest of your infrastructure. Settings supplied by the file are locked in the UI.

Once installed, sign in with the initial administrator account and follow the first-day checklist.

Upgrades

New versions of the platform and the Node Library are released as new images and chart versions. Upgrading is a regular Helm upgrade or Compose pull, and database migrations run automatically on startup. Flows keep working across upgrades: each placed node keeps the configuration it was created with until you choose to reload it.