Deployment overview

FastAPI requires ASGI. The right topology depends on how Django is deployed today.

Django already uses ASGI

Run one combined application:

Load balancer / Nginx
└── ASGI service
    ├── /api/* → FastAPI
    └── /*      → Django ASGI

Read the ASGI deployment guide.

Django currently uses WSGI

Keep the existing WSGI service and add an independently scalable ASGI service:

Load balancer / Nginx
├── /api/* → ASGI service → FastAPI + Django ASGI
└── /*      → WSGI service → Django WSGI

Read the WSGI deployment guide.

Invariants for both topologies

• Deploy exactly the same application version to every process.
• Run migrations once as a release job, not in every web container.
• Keep secrets in a secret manager, never an image or repository.
• Send correct proxy headers and trust only known proxy addresses.
• Size the total database pool across every worker and replica.
• Use a shared production session backend.
• Test graceful termination under real request duration.

The examples use Uvicorn and Gunicorn for portability. Daphne, uWSGI, or another conforming server can be substituted without changing the bridge.