Aviz Networks details integration of ONES with NVIDIA Spectrum-X for AI networking

Aviz Networks has integrated its Open Networking Enterprise Suite (ONES) with NVIDIA Spectrum-X to support Ethernet-based Graphics Processing Unit (GPU) cloud infrastructures, targeting optimized networking for Artificial Intelligence (AI) workloads. This integration aims to provide enterprise IT and security leaders with enhanced control over AI network orchestration and performance management.

Research overview

Aviz Networks’ ONES platform extends support for NVIDIA Spectrum-X, which combines Spectrum-4 switches with BlueField-3 SuperNICs to deliver lossless and high-performance networking tailored for AI Operations (AIOps). The ONES architecture employs agentless methods and containerized microservices to facilitate streamlined network planning, deployment, monitoring, and scalability within AI-driven environments.

By automating the provisioning of lossless fabrics and supplying comprehensive telemetry for congestion control, ONES assists in maximizing GPU utilization. This management is designed to provide predictable AI performance throughout the network fabric lifecycle, covering initial setup (Day 0) to ongoing operations (Day 2).

Key findings

Aviz Networks addresses challenges encountered during 5G deployment, including scattered monitoring systems and extensive data volume handling, along with cost containment pressures. The company’s Deep Network Observability solution delivers real-time insights with cohesive integration capabilities and scalable infrastructure management for modern mobile networks.

The solution utilizes optimized data ingestion and packet processing combined with an open, standards-based framework. This approach allows operators to detect network degradation swiftly—within five seconds—and precisely correlate Quality-of-Experience (QoE) and Service Level Agreement (SLA) metrics.

Technical breakdown

The Aviz solution reduces Deep Packet Inspection (DPI) hardware requirements and power consumption by 80%. It enables continuous monitoring of QoE and SLA compliance while offering lower Total Cost of Ownership (TCO) through adaptable hardware and extended upgrade intervals.

To support interoperability, the platform integrates with existing 4G, Long Term Evolution (LTE), and 5G NSA/SA networks, Cisco ACI fabric, and analytics pipelines via Kafka. It operates on standard x86 servers accelerated by Data Plane Development Kit (DPDK), consolidating traffic filtering, advanced packet processing, and real-time data export.

Leadership perspective

Thomas Scheibe, Chief Product Officer at Aviz Networks, highlighted the need for lossless, high-performance networking for AI at scale, stating that their collaboration with NVIDIA enables efficient planning, deployment, and monitoring of AI-ready fabrics. Gilad Shainer, NVIDIA’s Senior Vice President of Networking, noted that NVIDIA Spectrum-X assists providers like Aviz in delivering Ethernet-based multi-tenant AI networking that optimizes GPU performance while simplifying management.

Operational impact

The integration of Aviz ONES with NVIDIA Spectrum-X is being showcased at the NVIDIA GTC event, offering enterprises a demonstration of AI-driven network management capabilities. Customers such as Itochu Techno-Solutions Corporation report that the collaboration with Aviz has produced scalable, high-performance network observability solutions compatible with current infrastructures.

The platform consolidates packet processing and monitoring into a unified framework, reducing data silos and enabling faster, more informed network decisions, which are relevant for operators managing evolving AI and 5G network environments.

This Blog Signals brief provides a neutral, fact-based overview of Aviz Networks’ recent advancements in deep network observability and integration with NVIDIA Spectrum-X, delivering context for enterprise technology decision-makers navigating AI and 5G infrastructure challenges.