Skip to content

GCP Deployment

Prerequisites

  • GCP account with GPU quota
  • Terraform >= 1.0
  • gcloud CLI configured

Quick Deploy

bash
cd terraform/gcp

# Copy and edit variables
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars with your project ID

# Deploy
terraform init
terraform plan
terraform apply

Resources Created

  • GPU Instance: A100 GPU compute instance
  • VPC Network: Isolated network for Quila
  • Firewall Rules: Ports 8000, 8001 open
  • Storage Bucket: Model checkpoints and data

Configuration

Edit terraform.tfvars:

hcl
project_id   = "your-gcp-project-id"
region       = "us-central1"
zone         = "us-central1-a"
machine_type = "a2-highgpu-1g"
gpu_type     = "nvidia-tesla-a100"
gpu_count    = 1

Access

After deployment:

bash
# Get instance IP
terraform output instance_ip

# Access API
curl http://<instance_ip>:8000/status

Cleanup

bash
terraform destroy