Mycelium Runtime Library

What is Mycelium?

Mycelium is a framework that helps you easily write highly scalable (microservice) applications.

Mycelium automatically generates Terraform from your code, and handles load balancing for you.

The goal is to write your code without thinking about the infrastructure as if it was a simple monolithic application, and then have Mycelium decide on how to split your app into components, handle load balancing, communication between components, etc...

How to use this library?

Import this library into your app and use it to annotate your code with service hints/constraints.

The more this framework evolves, the less help you will have to give in the form annotations.

Finally, invoke the CLI (myc command) and watch it automatically generate Terraform for you (and optionally deploy it for you).

Example Usage

Create an empty Poetry application with myc (runtime) as dependency and myccli (compiler) as a development dependency.

Add some package (e.g. test_app) to your project with __init__.py containing:

from myc import Mycelium

app = Mycelium()
foo_service = app.create_service("foo")
bar_service = app.create_service("bar")

@foo_service.fastapi.get("/")
def foo():
    return f'The answer is: {bar()}'

@bar_service.fastapi.get("/")
def bar():
    return 42

Then run myc compile test_app:app to generate Terraform, followed by myc deploy to deploy your app!

What will happen:

• A Dockerfile will be generated for each service
• Docker images will be built and pushed to a private ECR repository
  • Temporary caveat: The framework expects you to have a repository called mycelium in your AWS region of choice (an update will come soon to have this repo created automatically)
• Two ECS services will be created in a new cluster for foo and bar services.
• The call to bar() in foo function will be transformed to an HTTP get request!

Limitations

This project is in its earliest stages, and feature count is limited at the moment. The framework (currently) assumes projects to use Poetry as a package manager

Currently supported features:

• AWS as a cloud provider
• ECS as a platform in AWS (Lambda support will come very soon)
• Poetry as the package manager

More will come soon!

Upcoming Features

• Dynamic and static load measurements to determine optimal load balancing strategy
• Support more cloud providers: Google Cloud, Azure, etc...
• Support more service platforms: AWS Lambda, EC2, etc...
• Support more package managers other than Poetry
• Flexible serialization/deserialization strategies
• Command to launch dashboard to watch:
  • Service topology
  • Aggregated logs
  • Load balancing metrics