architecture.md

SDA architecture

• _local: The local environement to launch for development, for example using flask's file detection and development server or CRA's webpack
• _prod: The production environement for the end build, building what's neccessary and

Stacks

1. Machine learning
   • local
     • Flask (with its development server)
   • production
     • Flask
     • Gunicorn
2. Science
   • local
     • Fast API
     • Uvicorn
   • production
     • Fast API
     • Uvicorn
3. Visualiser
   • local
     • React
     • Webpack/CRA
   • production
     • web bundle (served from nginx)

All load balanced/proxied via nginx

Access

1. Machine learning
   • host
     • localhost:8015/machine_learning/ (nginx)
     • localhost:5001 (direct)
   • docker
     • 0.0.0.0:5001
2. Science
   • host
     • localhost:8015/science/ (nginx)
     • localhost:5002 (direct)
   • docker
     • 0.0.0.0:8000
3. Visualiser
   • host
     • localhost:8015 (nginx)
     • localhost:3030 (direct) Preferable for development, the current nginx conf is not configured to work correctly with the websocket and with the subpath
   • docker
     • 0.0.0.0:5001

Environment Variables

• PATH_PREFIX The sub path on which the SDA is deployed on.

• SERVICE_INPUT_DATA (Read-Only) User's file space or the files passed into the SDA.

• SERVICE_APP_DATA (Read-Write-Execute) The app's appdata directory for persistent data or large datasets.

• SERVICE_USER_APP_DATA (Read-Write-Execute) The user's appdata directory, for string user inputs or other usefull peristent data.

• SERVICE_OUTPUT_DATA (Read-Write-Execute) The file space which will then be saved in the user's basket later on.