A cloud-native, open-source data platform for Kubernetes.
Modular, sovereign, and community-driven.
Get Started · Architecture · Community · Roadmap
OKDP is structured around two complementary layers:
- Data Platform: A curated catalog of battle-tested open-source data tools. Each component is independently deployable, with no lock-in to the Control Plane.
- Control Plane: A unified management layer (OKDP Server + OKDP UI) for deploying, configuring, and monitoring platform components across one or multiple clusters.
| Component | Description | Helm Chart | OKDP Repo |
|---|---|---|---|
| Apache Spark | Distributed data processing engine with Kubernetes-native operator, OIDC-enabled History Server, real-time Web Proxy, and pre-built Docker images | spark-operator / spark-history-server | spark-images · spark-history-server · spark-auth-filter · spark-web-proxy |
| Hive Metastore | Centralized metadata repository for data lakes, storing table schemas, partitions, and data locations for Spark, Trino, and other engines | hive-metastore | hive-metastore |
| Trino | Distributed SQL query engine for fast, interactive analytics across S3, Hive Metastore, and other data sources | trinodb/trino | upstream |
| Apache Superset | Enterprise BI platform for interactive dashboards, SQL Lab, and data visualization with OIDC authentication | okdp-superset | okdp-superset |
| Apache Airflow | Workflow orchestration platform for authoring, scheduling, and monitoring data pipelines and Spark jobs | apache/airflow | upstream |
| JupyterHub | Multi-user interactive notebook environment with PySpark integration, S3 file browsing, and OIDC authentication | jupyterhub | jupyterlab-docker |
| OKDP Examples | Hands-on examples, Jupyter notebooks, and data workflows showcasing the OKDP platform end-to-end | okdp-examples | okdp-examples |
For testing purpose, you can deploy all infrastructure prerequisites at once with the okdp-prerequisites umbrella chart. Do not use it for production cluster.
The OKDP Control Plane provides a unified interface to deploy, configure, and monitor all platform components across one or multiple Kubernetes clusters.
| Component | Description | Repo |
|---|---|---|
| OKDP Server | REST API backend for managing deployments, clusters, projects, GitOps repositories, and package catalogs | okdp-server |
| OKDP UI | Web interface for deploying and monitoring all OKDP services, with GitOps and direct Kubernetes deployment modes | okdp-ui |
The Control Plane is optional. All Data Platform components can be deployed independently via Helm without it.
The fastest way to explore OKDP. Spins up a full, pre-configured platform on Kind with a single command.
Deploy components individually on your own cluster using the Helm-based install guides.
See the helm-handbook for install guides.
OKDP is supported by TOSIT (The Open Source I Trust), initiated by DGFiP, Orange, and other organizations. The goal is a sovereign, powerful, and fully open-source data stack accessible to everyone.
- 🌐 Website: okdp.io
- 📅 Weekly technical meeting: Every Wednesday at 10:00 CET
- 🐳 Docker images: Quay.io/okdp
- ☕ Java artifacts: Maven Central (io.okdp)
- 🤝 Contribute: okdp.io#community
See the official roadmap on okdp.io.