LaunchFlow Logo

Introduction

Get Started with LaunchFlow

LaunchFlow is an open source Python SDK that lets you deploy websites, APIs, and other applications to AWS / GCP with minimal configuration. See the welcome page for an overview.

Follow this guide to launch a FastAPI app to ECS Fargate on AWS or Cloud Run on GCP.

1. Install LaunchFlow

Install the LaunchFlow Python SDK and CLI using pip.

1
pip install "launchflow[aws]"

2. Setup your local environment

Initialize LaunchFlow

Use the CLI to set up your local workspace.

1
lf init --backend=local

Running the command will create a launchflow.yaml file that looks like this:

1
project: your-project-name

2
backend: file://.launchflow

The --backend=local flag tells LaunchFlow to store configuration and state information locally. You can also use a cloud storage like Google Cloud Storage, S3 (coming soon), or LaunchFlow Cloud.

You can learn more about it in the launchflow.yaml docs.

Authenticate with Cloud Providers

LaunchFlow uses your local AWS credentials to manage and provision resources. We recommend using the AWS CLI to authenticate with your AWS account. Once you have the CLI set up, run the following command to make sure you're authenticated:

1
aws sts get-caller-identity

3. Launch Your Application

Create a FastAPI Application

Let's start out by creating a simple FastAPI application.

main.py:

1
from fastapi import FastAPI

2


3
app = FastAPI()

4


5
@app.get("/")

6
def index():

7
    return "Hello World"

To run the application locally, we need to install some dependencies:

1
pip install fastapi uvicorn

Then run it with:

1
uvicorn main:app --port 8080

If everything is set up correctly, you should see Hello World when you visit http://localhost:8080/ in your browser.

Deploy Your Application

Now that we have a working app, let's deploy it to the cloud!

First we'll need a Dockerfile to tell LaunchFlow how to build your application.

Dockerfile:

1
FROM public.ecr.aws/docker/library/python:3.11-slim

2


3
WORKDIR /code

4


5
RUN apt-get update && apt-get install -y --no-install-recommends gcc libpq-dev && apt-get clean && rm -rf /var/lib/apt/lists/*

6


7
RUN pip install --no-cache-dir -U pip setuptools wheel && pip install --no-cache-dir launchflow[aws] fastapi uvicorn

8


9
COPY ./main.py /code/main.py

10


11
ENV PORT=80

12


13
EXPOSE $PORT

14


15
CMD uvicorn main:app --host 0.0.0.0 --port $PORT

The only thing left to do is add the ECSFargate Service somewhere in your project directory. LaunchFlow will handle networking, security, and deployment configuration for you.

1
from fastapi import FastAPI

2
import launchflow as lf

3


4
ecs_fargate = lf.aws.ECSFargate("my-ecs-fargate")

5


6
app = FastAPI()

7


8
@app.get("/")

9
def index():

10
    return "Hello World"

We recommend using a single infra.py file to define your infrastructure, but you are free to use LaunchFlow anywhere in your project.

We're using a single file here for simplicity.

LaunchFlow will set up a Fargate Service and Docker build pipeline for you when you run:

1
lf deploy my-env

You will be able to view the plan and confirm before the resources are created. Deploy Plan

If everything goes well, you should see a link to your deployed service and logs to debug any issues that may have occurred.

Deploy Result

That's all there is to it!

Learn more about the deploy command here.

Add Cloud Resources

LaunchFlow makes it easy to add databases, task queues, and other cloud resources to your application. Your Services automatically have access to any Resources you create alongside them in the same Environment.

To add storage bucket to your application, simply import a LaunchFlow S3Bucket resource anywhere in your Python code. You can even use it inside your application code if you want!

main.py:

1
from fastapi import FastAPI

2
import launchflow as lf

3


4
my_service = lf.aws.ECSFargate("my-ecs-fargate")

5


6
# NOTE: The bucket name must be globally unique

7
bucket = lf.aws.S3Bucket(f"my-get-started-bucket-{lf.environment}")

8


9
app = FastAPI()

10


11
@app.get("/")

12
def index():

13
    # Read / Write to the bucket to test it out 

14
    bucket.upload_from_string("Hello World", "hello_world.txt")

15
    contents = bucket.download_file("hello_world.txt")

16
    return contents

Rerun the lf deploy command to create the S3Bucket resource and update the ECSFargate service with the new code.

1
lf deploy my-env

You can also run your application locally with the lf run command:

1
lf run my-env -- uvicorn main:app --port 8080

Visit http://localhost:8080/ to see the updated response!

Switch Environments

One of the main benefits of LaunchFlow is the ability to easily replicate your application across multiple environments. To create a staging version of your application, just run lf deploy again with a different environment name:

1
lf deploy my-staging-env

You can easily customize dynamic configuration across environments using the lf.environment global variable in your code.

1


2
force_destroy = True

3
desired_count = 1

4


5
# Set different configurations based on the environment

6
if lf.environment == "production":

7
    force_destroy = False

8
    desired_count = 2

9


10
bucket = lf.aws.S3Bucket("my-bucket", force_destroy=force_destroy)

11
api = lf.aws.ECSFargate("my-ecs-fargate", desired_count=desired_count)

See the GitHub Integration to learn how to automate deployments between environments on git push.

4. Visualize, Share, and Automate

LaunchFlow Console

LaunchFlow Cloud usage is optional and free for individuals.

Using the local backend like we did above works fine for starting a project, but doesn't offer a way to share state between multiple users. LaunchFlow Cloud is a web-based service for managing, sharing, and automating your infrastructure. It's free for individuals and provides a simple, secure way to collaborate with your team and automate your release pipelines.

Sign up for LaunchFlow Cloud and connect your local environment by running:

1
lf init --backend=lf

This will create a project in your LaunchFlow Cloud account and migrate your local state to the LaunchFlow Cloud backend.

What's next?

Previous
Welcome
Next
Why LaunchFlow