Installation (GCP)

Deploy Saturn Cloud Enterprise on Google Cloud Platform infrastructure

Saturn Cloud Enterprise on Google Cloud Platform (GCP) provides a managed data science and machine learning environment directly in your own GCP account. We offer scalable compute, a broad GPU portfolio, and native integration with BigQuery, Cloud Storage, and other GCP services.

Key Benefits

  • Access to NVIDIA A100 (40GB/80GB), L4, T4, and V100 GPUs
  • Managed Kubernetes with auto-scaling and monitoring
  • Regional deployments with up to 99.95% SLA for GKE control planes
  • Integration with native GCP services

How It Works (Architecture at a Glance)

Saturn Cloud on GCP

We provision a Google Kubernetes Engine (GKE) cluster, create a dedicated VPC, and configure CPU and GPU node pools. This setup delivers a secure, isolated, and scalable environment for AI workloads.

Core Components

  • GKE cluster with Workload Identity enabled
  • VPC network with custom subnets
  • CPU and GPU node pools tuned for different workloads
  • Google Cloud Storage integration
  • Built-in logging and monitoring

Deployment Model

Networking

  • Custom VPC {cluster-name}-vpc with manually created subnets
  • Primary subnet primary-subnet default CIDR 10.0.0.0/16
  • Network isolation with explicit topology control
  • Zonal or regional clusters supported

Load Balancer Options

  • Internal load balancer keeps traffic private inside your VPC
  • External load balancer provides public IPs, can be restricted with Cloud Armor, IP allowlists, or Identity-Aware Proxy

Node Pools

  • System nodes run the control plane
  • CPU node pools support N2 Standard, High Memory, Compute Optimized, and Memory Optimized instances
  • GPU node pools support NVIDIA A100, L4, T4, and V100 GPUs

For custom GPU types, autoscaling limits, or region availability, contact support@saturncloud.io.

Security and Compliance

  • We enable Workload Identity for secure pod-to-service authentication
  • We use Shielded nodes to protect against rootkits
  • We apply network policies for segmentation and fine-grained access
  • You can optionally deploy as a private cluster
  • We integrate Cloud IAM for unified access control

Storage and Data Access

  • We install the Persistent Disk CSI driver for dynamic volumes
  • We configure the GCS FUSE CSI driver for direct Cloud Storage access
  • We support regional or zonal deployment options for availability

Monitoring and Operations

  • We provide Cloud Logging and Monitoring dashboards
  • We enable cluster autoscaling and auto-repair
  • We set up built-in alerts and performance optimization

Installation Steps

  1. Create a dedicated GCP project for Saturn Cloud.

  2. Grant service account access:

    gcloud projects add-iam-policy-binding YOUR_PROJECT_ID \
  --member="serviceAccount:sce-installer@sce-gcp-manager.iam.gserviceaccount.com" \
  --role="roles/owner"

  3. Contact Saturn Support. Our team will deploy the infrastructure.

  4. Receive your Saturn Cloud URL and admin credentials.

Scaling and Ongoing Management

  • We provide automatic updates and patches
  • We offer 24/7 monitoring and support
  • We implement backup and disaster recovery
  • We handle security and performance tuning
  • Nodes scale to zero when idle
  • Optional preemptible VMs for savings
  • Resource tagging for cost tracking
  • Idle detection to shut down unused resources

GPU quota on GCP is limited. Contact us early to secure allocations for high-performance GPUs.

Regional Availability

We can deploy Saturn Cloud in any region supporting GKE and the required GPU types, such as:

  • us-central1 (Iowa)
  • us-east1 (South Carolina)
  • us-west1 (Oregon)
  • europe-west4 (Netherlands)
  • asia-southeast1 (Singapore)

Enterprise Advantages (Summary)

  • Broad GPU access including A100, L4, T4, and V100
  • High availability across many GCP regions
  • Deep integration with GCP services
  • Secure by design with enterprise-grade features