deploying-streamlit-to-snowflake

name	deploying-streamlit-to-snowflake
description	Deploy Streamlit applications to Snowflake using the `snow` CLI tool. Covers snowflake.yml manifest configuration, compute pools, container runtime, and deployment workflow.

Deploying Streamlit to Snowflake

Deploy local Streamlit apps to Snowflake using the snow CLI.

Prerequisites

• Snowflake CLI v3.14.0+: Run snow --version to verify
• A Streamlit app: Your main entry point file (e.g., streamlit_app.py)
• A configured Snowflake connection: Run snow connection list to verify

Deployment Workflow

Step 1: Create Project Structure

my_streamlit_app/
  snowflake.yml        # Deployment manifest (required)
  streamlit_app.py     # Main entry point
  pyproject.toml       # Python dependencies
  src/                 # Additional modules
    helpers.py
  data/                # Data files
    sample.csv

Quick start with templates:

# Single-page app
snow init my_app --template streamlit_vnext_single_page

# Multi-page app
snow init my_app --template streamlit_vnext_multi_page

Step 2: Create snowflake.yml

definition_version: 2
entities:
  my_streamlit:
    type: streamlit
    identifier:
      name: MY_APP_NAME           # App name in Snowflake
      database: MY_DATABASE
      schema: MY_SCHEMA
    query_warehouse: MY_WAREHOUSE
    compute_pool: STREAMLIT_DEDICATED_POOL
    runtime_name: SYSTEM$ST_CONTAINER_RUNTIME_PY3_11
    external_access_integrations:
      - PYPI_ACCESS_INTEGRATION   # For pip installs
    main_file: streamlit_app.py
    artifacts:
      - streamlit_app.py
      - pyproject.toml
      - src/helpers.py            # Include ALL files your app needs
      - data/sample.csv

Step 3: Deploy

cd my_streamlit_app
snow streamlit deploy --replace

The --replace flag updates an existing app with the same name.

Step 4: Access Your App

After deployment, snow outputs the app URL. You can also find it in Snowsight under Projects > Streamlit.

Configuration Reference

Parameter	Description	Example
name	Unique app identifier	MY_DASHBOARD
database	Target database	ANALYTICS_DB
schema	Target schema	DASHBOARDS
query_warehouse	Warehouse for SQL queries	COMPUTE_WH
compute_pool	Pool running the Python runtime	STREAMLIT_DEDICATED_POOL
runtime_name	Container runtime version	SYSTEM$ST_CONTAINER_RUNTIME_PY3_11
main_file	Entry point script	streamlit_app.py
artifacts	All files to upload (must include main_file)	See example above
external_access_integrations	Network access for pip, APIs	PYPI_ACCESS_INTEGRATION

Key Points

1. Always use container runtime (runtime_name) for best performance
2. List ALL files in artifacts - anything not listed won't be deployed
3. Dependencies go in pyproject.toml - installed automatically on deploy
4. Iterate with --replace - redeploy without creating duplicates

Troubleshooting

App not updating?

• Ensure you're using --replace
• Check that changed files are in artifacts

Import errors?

• Verify all modules are in artifacts
• Check pyproject.toml has all pip dependencies

Network/pip errors?

• Add PYPI_ACCESS_INTEGRATION to external_access_integrations