Cloud Integration Guide

AWS Integration

Mount S3 Buckets

Access your S3 data directly from GPU instances:

# Install s3fs
sudo apt install s3fs

# Configure credentials
echo ACCESS_KEY:SECRET_KEY > ~/.passwd-s3fs
chmod 600 ~/.passwd-s3fs

# Mount bucket
s3fs my-bucket /mnt/s3 -o passwd_file=~/.passwd-s3fs

# Access data
ls /mnt/s3/datasets/

VPN to AWS VPC

Connect TerraGPU to your AWS VPC:

1. Create VPN Gateway in your VPC
2. Create Customer Gateway pointing to TerraGPU (IP provided by support)
3. Create VPN Connection
4. Download configuration and send to support@terragpu.com
5. Salt Labs configures connection (15-30 minutes)

Azure Integration

Mount Azure Blob Storage

# Install blobfuse
wget https://packages.microsoft.com/config/ubuntu/22.04/packages-microsoft-prod.deb
sudo dpkg -i packages-microsoft-prod.deb
sudo apt update
sudo apt install blobfuse

# Configure
echo "accountName myaccount" > ~/fuse_connection.cfg
echo "accountKey KEY" >> ~/fuse_connection.cfg
chmod 600 ~/fuse_connection.cfg

# Mount
sudo mkdir /mnt/azure
blobfuse /mnt/azure --tmp-path=/tmp/blobfuse --config-file=~/fuse_connection.cfg

Google Cloud Integration

Mount GCS Buckets

# Install gcsfuse
export GCSFUSE_REPO=gcsfuse-`lsb_release -c -s`
echo "deb https://packages.cloud.google.com/apt $GCSFUSE_REPO main" | sudo tee /etc/apt/sources.list.d/gcsfuse.list
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
sudo apt update
sudo apt install gcsfuse

# Authenticate
gcloud auth application-default login

# Mount
gcsfuse my-bucket /mnt/gcs

MLOps Tools

Weights & Biases

pip install wandb
wandb login  # Enter API key

# In your training script
import wandb
wandb.init(project="my-project")

MLflow

pip install mlflow

# Track experiments
import mlflow
mlflow.set_tracking_uri("http://your-mlflow-server")
mlflow.start_run()
mlflow.log_param("learning_rate", 0.001)
mlflow.log_metric("accuracy", 0.95)

Hugging Face

pip install transformers datasets
huggingface-cli login  # Enter token

# Load models and datasets
from transformers import AutoModel
model = AutoModel.from_pretrained("bert-base-uncased")

CI/CD Integration

GitHub Actions

name: Train Model
on: [push]
jobs:
  train:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v2
      - name: Launch TerraGPU Instance
        run: |
          curl -X POST https://api.terragpu.com/instances \
            -H "Authorization: Bearer ${{ secrets.TERRAGPU_API_KEY }}" \
            -d '{"instanceType":"a100-40gb","name":"ci-training"}'
      - name: Run Training
        run: ssh ubuntu@instance-ip python train.py

Need Help?

Email: support@terragpu.com