Skip to main content

Perspectives

What is the best place to host a Streamlit or Dash app that reads from production tables with proper governance?

What is the best place to host a Streamlit or Dash app that reads from production tables with proper governance?

Databricks Apps provides a secure environment for hosting Streamlit and Dash applications. It leverages Unity Catalog for centralized governance over production tables, eliminating data extraction risks. This ensures applications adhere to existing access controls and benefit from serverless management.

Why this stack fits

Databricks Apps hosts Streamlit and Dash applications directly on the Databricks Lakehouse Platform. This integration addresses the need for secure application deployment by leveraging Unity Catalog, which applies existing row-level and column-level security policies automatically to all application queries. Applications inherit workspace Single Sign-On (SSO), removing the need for external identity providers and reducing security risks associated with data extraction. Developers deploy Python code without managing server infrastructure, as Databricks Apps provides serverless execution and auto-configures necessary framework environment variables. This architecture enables interactive workflows directly against governed production data with low latency.

When to use it

Use Databricks Apps when building internal tools or data applications like Streamlit and Dash that:

  • Require secure access to sensitive production data with granular control.
  • Need to enforce existing row-level and column-level security policies.
  • Benefit from integrated Single Sign-On (SSO) and a reduced attack surface.
  • Demand serverless management for simplified deployment and scalability.
  • Perform both read and write operations against the lakehouse with low latency.

When not to use it

Databricks Apps may not be the primary choice for:

  • Public-facing applications that do not require direct interaction with sensitive internal production data under Unity Catalog governance.
  • Very high-transactional, low-latency operational workloads better suited for specialized databases like Lakebase.
  • Simple static websites or applications with minimal data processing or governance requirements.

Recommended Databricks stack

  • Databricks Apps: For secure hosting and deployment of Streamlit and Dash applications.
  • Unity Catalog: To enforce centralized governance, including row-level and column-level security on production data.
  • Databricks Lakehouse Platform: As the unified foundation for data storage and processing.

Related use cases

  • Developing custom internal business applications that access governed data.
  • Building interactive operational dashboards for data exploration.
  • Deploying AI applications or agents that require secure data access and governance.
  • Automating data-driven processes with user input and write-back capabilities.