MultiCloud Deployment Project

This project demonstrates a unified codebase for deploying a Python-based function to both AWS Lambda and Azure Functions. The solution is structured to isolate business logic from cloud integration code.

The objective is to show how a Cloud Agnostic solution can utilize Cloud Native services from both Azure and AWS, while keeping implementation differences in the CI/CD pipeline instead of the application code.

Overview

• Business Logic (business_logic.py): Contains the core functionality, such as processing records or calculations.
• Abstract Interfaces (interfaces.py): Defines interfaces that decouple business logic from cloud-specific implementations.
• AWS Implementations (azure_impl.py): Uses AWS services (e.g., DynamoDB, S3, Lambda).
• Azure Implementations (aws_impl.py): Uses Azure services (e.g., CosmosDB, Blob Storage, Functions).
• Unified Handler (normalize_handler.py): A common entry point that normalizes input and routes execution to the correct business logic.

Key Features:

• Uses Azure services:
  • Azure Function Apps
  • CosmosDB
• Uses AWS services:
  • API Gateway
  • AWS Lambda
  • DynamoDB
• Implements a multi-cloud strategy without tightly coupling the application to a single provider.
• Ensures code portability, with cloud-specific differences handled in the deployment pipeline.

Architecture

The project follows a cloud-agnostic approach, where:

• Business logic remains consistent across different cloud providers.
• Cloud-specific implementations are managed via infrastructure as code (IaC) and CI/CD pipelines.

Getting Started

Prerequisites

To run this project, ensure you have the following installed:

• Python
• Azure CLI and AWS CLI (for cloud interactions)
• Git for version control

Deployment

AWS Lambda

• Packaging: The Makefile creates a zip package that includes a top-level folder (e.g., package) so that AWS Lambda can import modules with absolute imports.
• Handler Configuration: Set the Lambda handler to package.main.lambda_handler.

Azure Functions

• Packaging: The Makefile also creates a deployment package for Azure with a proper structure including host.json, requirements.txt, .python_packages, and the MyFunction folder.
• Handler: The function’s init.py in MyFunction uses relative imports to load unified_handler from main.py.

GitHub Actions

The project uses GitHub Actions for automated deployment. The workflow includes:

• Building and packaging the code for both AWS and Azure.
• Deploying to AWS Lambda (using the AWS CLI).
• Deploying to Azure Functions (using the Azure/functions-action with scm-do-build-during-deployment: true).

Configuration

• Environment Variables:
  • For AWS: Configure environment variables such as DYNAMODB_TABLE.
  • For Azure: Configure settings like COSMOSDB_ENDPOINT, COSMOSDB_KEY, AZURE_FUNCTION_APP_NAME, etc. Use a local.settings.json file for local development (do not deploy it).
• Secrets: Store sensitive information (e.g., AWS keys, Azure credentials) in GitHub Secrets.

Running Locally

To test locally:

1. For Azure Functions, place local.settings.json in the project root (next to host.json) and run: func start
2. For AWS Lambda, simulate the package structure by running: python -m package.main

License

MIT License