FastApi With AWS Serverless powered by CDK Typescript

Abstract

Deploy FastAPI in a Lambda function that is fronted by an HTTP API in API Gateway, you can enable API key required for the API

Open Table Of Contents

🚀 Solution overview
🚀 Build FastAPI as a lambda funnction handler
🚀 Deploy
🚀 Test API
🚀 Conclusion

🚀 Solution overview

Use AWS APIGW to build a simple API server with lambda integration
APIGW route paths such as /api/v1/ and /chat_gpt/ require API key (with usage plan)
The lambda function contains FastApi code to serves API requests and responses.
Let’s see the stack relationships

<img src=https://raw.githubusercontent.com/vumdao/simple-serverless-fastapi/master/images/diagram.png width=300>

🚀 Build FastAPI as a lambda funnction handler

Lambda handler source code

➜  simple-serverless-fastapi tree src/lambda-handler
src/lambda-handler
├── api
│   └── api_v1
│       ├── api.py
│       └── endpoints
│           └── users.py
├── main.py
└── requirements.txt

4 directories, 4 files

Direct route paths include / and chat_gpt

@app.get("/")
async def root():
    return {"message": "FastAPI running in a Lambda function"}


@app.get("/chat_gpt/")
async def read_chatgpt(question: str = None):
    return {"message": f"We got question: {question}"}

Restructure FastAPI Routing for developing API and optimize source code by using APIRouter

src/lambda-handler/api
└── api_v1
    ├── api.py
    └── endpoints
        └── users.py

from api.api_v1.api import router as api_router
app.include_router(api_router, prefix="/api/v1")

For the lambda function handler, we use Mangum python module which is an adapter for running ASGI applications in AWS Lambda to handle Function URL, API Gateway, ALB, and Lambda@Edge events.
For building API Docs, one must set the following parameters in FastApi() constructor to resolve /openapi.json correctly
- For API URL using APIGW stage URL, set root_path equal to the API stage name, eg. root_path=/AppAPI
- For API custom domain
```
docs_url='/docs',
openapi_url='/openapi.json',
```

🚀 Deploy

For production, building CDK pipeline for this is the best practice.

For the demo, I run cdk deploy manually

➜  simple-serverless-fastapi cdk deploy
 ✅  SimpleFastApiServerless

✨  Deployment time: 72.44s

The API GW and method request

<img src=https://raw.githubusercontent.com/vumdao/simple-serverless-fastapi/master/images/apigw.png width=1100>

<img src=https://raw.githubusercontent.com/vumdao/simple-serverless-fastapi/master/images/chat_gpt.png width=1100>
Custom domain mapped to the API

<img src=https://raw.githubusercontent.com/vumdao/simple-serverless-fastapi/master/images/custom-domain.png width=1100>

🚀 Test API

Open API Docs

<img src=https://raw.githubusercontent.com/vumdao/simple-serverless-fastapi/master/images/api_docs.png width=1100>

Call /chat_gpt with API key and query question

➜  simple-serverless-fastapi curl -X GET -H "Content-Type: application/json" -H 'x-api-key: 6sUnYj8PAw8MKu8O6FqSw1kf1clmC0Fx8ilQhVeO' https://chatgpt.simflexcloud.com/chat_gpt/ -d 'question=how%20are%20you' -G
{"message":"We got question: how are you"}

🚀 Conclusion

We created a FastAPI application using AWS Serverless. The user must provide the API key to query request and the API key is associated with the usage plan where we can specify who can access the deployed API stages and methods, and optionally sets the target request rate to start throttling requests.

References:

Simple Serverless FastAPI with AWS Lambda