
Deploy a Self-Hosted Data Lakehouse in Under an Hour
Lakehouse Studio Team
Deploy a Self-Hosted Data Lakehouse in Under an Hour
If you've been considering moving off Databricks—or you're a data team that never wanted a cloud vendor in the first place—this post walks through getting Lakehouse Studio running from scratch.
Prerequisites
A Linux VM or bare metal server (4 vCPUs, 16 GB RAM minimum)
Docker and Docker Compose v2 installed
Port 3000 accessible from your workstation
Step 1 — Clone and Run Setup
git clone <repository>
cd lakehouse-studio
bash infra/setup.sh
The setup wizard prompts for your admin email, password, and optional SMTP configuration, then writes a .env file.
Step 2 — Start the Stack
docker compose up -d
This starts seven containers:
PostgreSQL
Apache Spark master
Two Spark workers
Unity Catalog
Lakehouse Studio web application
Spark Thrift Server
Step 3 — Log In
Open:
http://your-server:3000
Sign in using the admin credentials you created during setup.
Step 4 — Connect Power BI
Navigate to Admin → BI Connections.
Copy the auto-generated connection string.
In Power BI Desktop, select Get Data → Spark.
Paste the host and port from the connection string.
Your Unity Catalog tables should appear in the schema navigator within seconds.
Published by
Datrixen Analytica