GREENACCOUNTER

Table of Contents

• Overview
• Getting Started
  • Prerequisites
  • Installation
• Features
• Project Structure
  • Project Index

Overview

GreenAccounter is an all-in-one developer toolset for energy and carbon footprint monitoring within distributed systems. It integrates backend services, deep learning models, and frontend interfaces to deliver real-time environmental insights.

Why GreenAccounter?

This project aims to facilitate energy-efficient and environmentally conscious system management. The core features include:

• 🌐 Globe: Multi-region electricity data collection and analysis for comprehensive environmental insights.
• 🛠️ Gear: Automated deployment pipelines using Docker and Kubernetes for reliable, scalable infrastructure.
• 🔒 Lock: Secure remote server management via SSH, ensuring trusted operations across cloud environments.
• 📈 Chart: Real-time visualization of resource utilization and carbon emissions to support proactive decision-making.
• Performance: Utilities for monitoring and optimizing system performance and responsiveness.

Features

Project Structure

└── GreenAccounter/
    ├── GreenAccounter-backend-connection
    ├── GreenAccounter-backend-monitor-carbon
    ├── GreenAccounter-backend-monitor-server
    ├── GreenAccounter-backend-orchestrator
    ├── GreenAccounter-frontend-gateway
    ├── LICENSE
    ├── README.md
    ├── config.py
    ├── firebase-adminsdk-fbsvc.json
    ├── make_file.bash
    ├── .env
    └── ssh_data.csv

Project Index

Getting Started

Prerequisites

Before deploying GreenAccounter, ensure you have the following installed and configured:

• Kubernetes Cluster: v1.20 or higher
• Docker: For building and managing container images

Installation

Build GreenAccounter from the source and install dependencies:

1. Clone the repository:

    ❯ git clone https://github.com/jhparkland/GreenAccounter
   

2. Navigate to the project directory:

    ❯ cd GreenAccounter
   

Configuration

1. Firebase Configuration

Edit a file named config.py in the project root directory and add the following information:

• Firebase AUTH key(JSON file name)
• Firebase Storage Bucket

Download the Firebase Admin SDK JSON key file from the Firebase Console and place it in the project root directory as firebase-adminsdk-fbsvc.json.

2. ElectricityMaps API Key

Create a file named .env and add your ElectricityMaps API key:

ELECTRICITYMAPS_API_KEY=<YOUR_ELECTRICMAPS_API_KEY>

3. SSH Configuration

Configure SSH credentials in ssh_data.csv for remote server access:

[Example: ssh_data.csv]

Place your RSA private keys in the respective rsa_keys/ directories within each backend module.

Deployment

Automated Kubernetes Deployment

After completing all configuration steps, deploy the entire project to Kubernetes:

bash make_file.bash

This script will:

• Copy configuration files to all backend services
• Build Docker images for frontend and backend components
• Push images to your container registry
• Deploy all services to your Kubernetes cluster
• Configure NGINX Ingress for routing

Manual Deployment (Optional)

If you prefer manual deployment, follow these steps:

1. Build Docker images:

   # Frontend
   cd GreenAccounter-frontend-gateway
   docker build -t greenaccounter-frontend:latest .
      
   # Backend services
   cd ../GreenAccounter-backend-connection
   docker build -t greenaccounter-connection:latest .
      
   cd ../GreenAccounter-backend-orchestrator
   docker build -t greenaccounter-orchestrator:latest .
      
   cd ../GreenAccounter-backend-monitor-server
   docker build -t greenaccounter-monitor-server:latest .
      
   cd ../GreenAccounter-backend-monitor-carbon
   docker build -t greenaccounter-monitor-carbon:latest .
   

2. Deploy to Kubernetes:

   # Deploy backend services
   kubectl apply -f GreenAccounter-backend-connection/back.yaml
   kubectl apply -f GreenAccounter-backend-orchestrator/back.yaml
   kubectl apply -f GreenAccounter-backend-monitor-server/back.yaml
   kubectl apply -f GreenAccounter-backend-monitor-carbon/back.yaml
      
   # Deploy frontend and ingress
   kubectl apply -f GreenAccounter-frontend-gateway/front.yaml
   

3. Verify deployment:

   kubectl get pods
   kubectl get services
   kubectl get ingress
   

Deep Learning Model Setup

Configuration

Before deploying the deep learning model, configure the following files in the VGGNET/ directory:

1. Firebase Configuration

Edit a file named config.py in the project root directory and add the following information:

• Firebase AUTH key(JSON file name)
• Firebase Storage Bucket

Download the Firebase Admin SDK JSON key file from the Firebase Console and place it in the project root directory as firebase-adminsdk-fbsvc.json.

2. ElectricityMaps API Key

Create a file named .env and add your ElectricityMaps API key:

ELECTRICITYMAPS_API_KEY=<YOUR_ELECTRICMAPS_API_KEY>

3. Carbon Configuration

Configure carbon thresholds in the carbon_config.csv file for carbon-aware learning:

[Example: carbon_config.csv]

Note: In this demo version, regions are pre-configured for testing purposes.

Docker Deployment for Deep Learning Models

The VGGNet deep learning model is containerized and deployed in cloud environments.

Note: This demo version uses VGGNet as a sample model for testing purposes.

Deployment Process:

1. Build the Docker image locally:

   cd VGGNET
   docker build -t vggnet .
   

2. Deploy to each region:

   # Run containers image to KR server
   $ user@kr-server "docker run --gpus all --name KR -it vggnet"
      
   # Run containers image to US server
   $ user@us-server "docker run --gpus all --name US -it vggnet"
      
   # Run containers image to IT-CSO server
   $ user@it-server " docker run --gpus all --name IT-CSO -it vggnet"
   

   Important Notes:

   • Container names must match the region codes (KR, US, IT-CSO)
   • --gpus all flag enables GPU access for model training
   • Ensure NVIDIA Docker runtime is installed on each server
   • Images are deployed locally without using a container registry

Each regional instance operates independently, processing region-specific electricity data and carbon intensity metrics.

4. Starting Training

Launch deep learning training in the terminal of your desired cloud server.

For example, to train the VGGNet model, use the following command:

$ docker run -it --gpus all vggnet --epoch 100 --lr 0.001 --batch 8 --vgg_model VGG16 --cuda 0 --step_size 30 --gamma 0.1 --resumption 0 --ssh_server 0

Command Options:

• --epoch: Number of training epochs
• --lr: Learning rate
• --batch: Batch size
• --vgg_model: VGGNet model variant to use
• --cuda: GPU device number
• --step_size: Step size for learning rate decay
• --gamma: Learning rate decay factor
• --resumption: Migration resumption flag (0: new training, 1: resume)
• --ssh_server: Current training SSH server

Each option can be modified according to your deep learning model requirements.

Each regional instance operates independently, processing region-specific electricity data and carbon intensity metrics.

Accessing the Application

Once deployed, access the GreenAccounter dashboard:

http://localhost/

Troubleshooting

Common Issues:

• Pods not starting: Check logs with kubectl logs <pod-name>
• Connection errors: Verify SSH credentials in ssh_data.csv
• API errors: Ensure ElectricityMaps API key is valid
• Firebase errors: Verify Firebase configuration and credentials

🎥 Demo Video

Watch the demo in action!

Check out the demo!

This repository is intended for demonstration purposes only.