Installation (Oracle)

Deploy Saturn Cloud Enterprise on Oracle Cloud Infrastructure

Saturn Cloud Enterprise on Oracle Cloud Infrastructure (OCI) delivers a fully managed AI and data science platform using Oracle’s high-performance infrastructure. We provide direct access to bare metal A100 and V100 GPU clusters, enterprise-grade security, and predictable performance for large-scale machine learning workloads.

Key Benefits

  • Direct access to bare metal A100 (40GB/80GB), V100, A10, L40S, and H100 for high-performance computing and AI workloads.
  • Managed Kubernetes with Oracle Kubernetes Engine (OKE) and built-in autoscaling
  • High-speed RDMA networking for distributed training at scale
  • Private networking options and integration with OCI IAM for secure access
  • Competitive pricing and capacity reservations for predictable cost and performance

How It Works (Architecture at a Glance)

Saturn Cloud on OCI

We deploy into your Oracle Kubernetes Engine (OKE) cluster inside your OCI tenancy, create a dedicated Virtual Cloud Network (VCN) with public and private subnets, and configure specialized node pools including bare metal A100 GPU instances. This setup gives your teams secure, isolated, and scalable compute power without virtualization overhead.

Core Components

  • OKE cluster with GPU scheduling
  • Dedicated VCN with custom subnets and advanced routing
  • Bare metal GPU clusters with NVIDIA NVLink and RDMA networking
  • OCI File Storage Service (FSS) and Object Storage integration
  • IAM and workload identity for secure authentication

Deployment Model

Networking

  • Custom VCN with configurable CIDR (default: 10.0.0.0/16)
  • Public subnets for load balancers and bastion hosts
  • Private subnets for worker nodes and cluster networks
  • Multiple availability domains for resilience
  • Service Gateway for private access to OCI services

Node Pools

  • System node pool runs Saturn Cloud control plane and system services
  • CPU node pools handle preprocessing, model serving, and general workloads
  • GPU node pools leverage bare metal A100 GPU shapes for large-scale training

Cluster configurations, GPU topologies, and network settings can be customized to meet performance and capacity requirements. Contact support@saturncloud.io for customization.

Security and Compliance

  • Integration with OCI IAM for fine-grained access control
  • Workload identity to map Kubernetes service accounts to OCI principals
  • Encrypted persistent storage at rest with OCI keys
  • Private networking and Network Security Groups
  • Comprehensive audit logging for compliance

Storage and Data Access

  • High-performance NFS storage via OCI File Storage Service (FSS)
  • Object Storage for scalable training data and model artifacts
  • Block Storage for databases and applications
  • Automated backup and recovery

Monitoring and Operations

  • OKE autoscaling with dynamic node provisioning
  • Cluster health monitoring and GPU performance analytics
  • Built-in alerts and optimization for distributed training
  • Continuous resource and cost monitoring

Installation Steps

  1. Prepare your OCI tenancy with the required service limits and compartments.

  2. Reserve bare metal GPU capacity for guaranteed availability.

  3. Provide credentials and infrastructure parameters to Saturn Cloud support:

    tenancy_ocid = "ocid1.tenancy.oc1..aaaaaaaaxxxxxxxxxx"
region = "us-ashburn-1"
compartment_id = "ocid1.compartment.oc1..aaaaaaaaxxxxxxxxxx"

vcn_cidr = "10.0.0.0/16"
public_subnet_cidrs = ["10.0.1.0/24", "10.0.2.0/24"]
private_subnet_cidrs = ["10.0.3.0/24", "10.0.4.0/24"]

cluster_name = "saturn-cluster-oci"
kubernetes_version = "v1.29.1"

  4. Grant Saturn Cloud support the necessary IAM policies.

  5. Our team deploys the OKE cluster and provisions the bare metal GPU infrastructure.

  6. You receive your Saturn Cloud URL and admin credentials.

Scaling and Ongoing Management

  • Automatic Kubernetes updates and GPU driver management
  • 24/7 monitoring and support for bare metal infrastructure
  • Performance and capacity planning
  • Hardware health monitoring and replacement
  • Backup and disaster recovery
  • Continuous cost optimization

Bare metal A100 capacity is limited. Contact us early to secure reservations and ensure rapid deployment of high-performance AI infrastructure.

Regional Availability

Saturn Cloud on OCI is available in regions where bare metal GPU instances are offered:

  • us-ashburn-1 (Virginia) – primary region with full A100 capacity
  • us-phoenix-1 (Arizona) – west coast option
  • Additional regions as OCI expands GPU infrastructure

Enterprise Advantages (Summary)

  • Bare metal performance with no virtualization overhead
  • High-speed interconnects for distributed training at scale
  • Predictable performance using dedicated resources
  • Enterprise reliability with Oracle’s infrastructure and SLAs
  • Cost-effective scaling through competitive pricing and per-second billing