Virtana is partnering with NVIDIA to advance observability for enterprise AI Factories. 

The integration brings together Virtana’s observability platform and NVIDIA’s AI and accelerated computing technologies to help IT teams manage complex, distributed AI environments more efficiently and reliably.

Virtana, an NVIDIA Connect program member, offers unified observability across on-premises, cloud, and containerized environments, supporting deeper observability for NVIDIA GPU-accelerated infrastructure to deliver faster insights, automation, and performance optimization.

“To accelerate Virtana’s mission to deliver AI Factory Observability, powered by AI, at industrial scale, our collaboration with NVIDIA is critical,” said Paul Appleby, CEO and President of Virtana. “By combining Virtana’s deep expertise in hybrid cloud performance with NVIDIA’s market-leading computing and AI capabilities, we’re empowering enterprises to improve application performance, accelerate root cause analysis, and reduce infrastructure costs. Our collaboration gives IT teams the intelligence they need to support AI-native workloads with confidence and efficiency.”

The collaboration is focused on delivering intelligent, real-time insights that support faster decision-making, resource efficiency, and improved application performance. With deeper observability into NVIDIA GPU-powered environments, enterprise IT teams can reduce mean-time-to-resolution (MTTR), align infrastructure with business performance and cost goals, and gain readiness for AI-native application deployments.

Key capabilities of the Virtana Platform include:

• Automated Topology Discovery: Leveraging machine learning, the platform can more rapidly map interdependencies between AI applications, GPUs, storage, and network components—enabling real-time visibility into system behavior and potential bottlenecks.
• AI-Based Root Cause Analysis: Virtana will use NVIDIA AI Enterprise to improve the speed and precision of root-cause identification, analyzing large datasets in seconds to minimize downtime and service impact.
• Predictive Performance Management: With predictive analytics powered by historical and real-time data, IT teams can address issues before they impact critical operations or user experience.
• Cost and Capacity Optimization: AI-driven insights support better forecasting, helping teams align resource usage with business needs while controlling costs for GPUs.
• Natural Language Query via Virtana Copilot: Virtana’s generative AI assistant, Copilot, allows users to query infrastructure data using natural language, making insights more accessible to non-technical users.
• Enhancing Observability for NVIDIA NIM with OpenTelemetry and Virtana: Virtana brings robust observability to NVIDIA NIM by leveraging OpenTelemetry standards. Our platform delivers deep visibility into application performance, health, and availability—empowering teams to monitor, trace, and optimize AI workloads running on NIM through OpenTelemetry.
• Enabling AI Infrastructure Observability: Virtana’s observability platform will provide visibility into the performance and behavior of AI workloads running on these systems, helping customers:

   Detect performance anomalies in real time.
   Assess the infrastructure impact of AI workloads.
   Plan for resource requirements in future AI deployments.
   Avoid downtime through proactive monitoring.

These capabilities are designed to support AI deployments in enterprise data centers, public clouds, and AI factories.

Virtana and NVIDIA continue to explore other opportunities to partner on enhanced observability solutions for AI workloads. Areas of future development include:

• Support for additional NVIDIA platforms, such as NVIDIA DGX systems.
• Integration with NIM (NVIDIA Inference Microservices) for performance optimization and efficient token cost management.
• Co-development of observability practices tailored for AI-driven enterprises.

This long-term collaboration aims to empower customers with the tools and intelligence needed to optimize infrastructure for modern workloads.