Optimizing Spectrum-X Deployments with OpenNebula

Operate secure, multi-tenant AI-ready infrastructure, fully integrated with NVIDIA Spectrum-X Ethernet networking.

Why OpenNebula on Spectrum-X Ethernet

OpenNebula integrates natively with Spectrum-X Ethernet to deliver a production-ready, software-defined cloud platform optimized for modern AI and accelerated workloads, enabling cloud operators to deploy and manage AI infrastructure with confidence.



End-to-End Multi-Tenancy

Isolated tenant environments are created on demand using VRFs, with strict separation across compute, GPU, and network layers while sharing the same Spectrum-X Ethernet fabric, supporting secure multi-tenant operation.



Native East–West Performance

Direct PCI passthrough of GPUs and high-performance SuperNICs enables line-rate east–west traffic between AI workloads, minimizing overhead and ensuring consistent performance for latency- and throughput-sensitive environments.



Controlled North–South Connectivity

Virtual machines attach seamlessly to north–south networks, allowing secure ingress and egress while preserving tenant isolation and traffic policies across shared cloud infrastructure.



Operational Simplicity at Scale

Network configuration, tenant provisioning, and device attachment are fully automated through OpenNebula, reducing manual intervention and operational risk in complex environments.

Main Features

OpenNebula delivers enterprise-grade cloud capabilities tailored for Spectrum-X-Ethernet-powered AI infrastructure, designed to support standardized and repeatable operations.

Spectrum-X Ethernet Fabric Automation

OpenNebula SDN integrates with the Spectrum-X Ethernet fabric by issuing NVUE commands directly to leaf switches, ensuring consistent and reproducible network state across deployments.

Lazy Tenant and VRF Provisioning

Tenants and VRFs are created only when needed, optimizing resource utilization and accelerating deployment as infrastructure requirements evolve.

GPU Passthrough

Dedicated GPUs and SuperNICs are assigned directly to tenant VMs, delivering native performance for AI and data-intensive workloads while maintaining clear hardware assignment.

Integrated N/S and E/W Network

Automated attachment of interfaces to tenant VRFs and north–south networks enables precise control over traffic flows across the infrastructure using centralized orchestration.

OpenNebula on NVIDIA Air

The OpenNebula control plane is already integrated and fully operational on NVIDIA Air. This provides a powerful, simplified, and cost-effective environment for running integration work and validating deployments at scale—particularly for large AI infrastructures such as AI Factories and AI Gigafactories, where repeatability and consistent testing are essential.

With NVIDIA Air, customers can get hands-on experience with OpenNebula and evaluate its integration with Spectrum-X Ethernet in a controlled environment. Beyond functional validation, Air also supports advanced capabilities such as network and system simulations (including digital twins), automation workflows, configuration rendering, and hardware evaluation, making it a practical platform for both assessment and large-scale proof points.

Go to NVIDIA Air

Learn More about OpenNebula with NVIDIA

OpenNebula Validated with NVIDIA Spectrum‑X Ethernet Networking

Discover how OpenNebula, integrated with NVIDIA Spectrum‑X Ethernet Networking, enables organizations to build high-performance, multi-tenant AI infrastructures that are flexible, efficient, and free from vendor lock-in.

Read the Press Release