Most teams don’t need another dashboard; they need answers that hold up in a post-incident review and a budget meeting. This release focuses on that outcome. It makes token spend, GPU efficiency, Kubernetes health, storage performance, and network behavior visible in one operating picture you can act on. You’ll see where dollars are going, why latency spikes happen, and which fixes reduce waste without risking reliability. Support for managed Kubernetes, deeper storage and backup visibility, and practical export and sharing options keep the workflow simple: observe, prove, fix, report.
Global View: Cost and performance clarity without tool switching for AI Workloads
Token Usage Dashboard
Most teams can’t see token consumption in the same place they track performance and errors. The Token Usage Dashboard connects token usage from LLM-as-a-Service to the metrics your business cares about: cost, latency, and error rates. This makes it possible to spot inefficient prompts or bursty workloads and tie them to real dollars and user impact.
Business value: Clear line of sight from usage to cost and reliability, so product and finance can agree on what to tune or throttle.
GPU Fleet Analysis Dashboard
GPU fleets are expensive to run and easy to underutilize. This dashboard shows power, temperature, utilization, and spend across on-prem and cloud, alongside indicators that point to throttling and waste. Root-cause assistance helps teams connect hot spots or idle periods to specific jobs and dependencies.
Business value: Reduce waste, prevent slowdowns from thermal and power issues, and defend budgets with evidence.
Service Observability alerts visible in Global View
Alerts from Virtana Service Observability now appear next to signals from Container Observability, Infrastructure Observability, and third-party sources. Teams can review incidents in one place instead of bouncing between consoles.
Business value: Faster triage and fewer missed handoffs, which improve MTTR.
These are just a few of the latest capabilities available to Virtana customers. Read the Global View technical blog to learn more about the rest.
Container Observability: Day-2 confidence for managed and hybrid Kubernetes
GKE Autopilot support and partnership
Virtana Container Observability is approved to run on GKE Autopilot. You get real-time performance visibility across managed clusters, plus the ability to cross-launch for deeper pod-level investigation when needed.
Business value: Operate on Google’s managed Kubernetes with confidence while keeping the same troubleshooting workflows.
AWS FSx for NetApp ONTAP monitoring
Storage is often the hidden cause of tail latency. You can now monitor FSx for ONTAP and correlate those signals with on-prem and cloud events through Global View’s correlation policies.
Business value: One place to see storage behavior across environments, which speeds up incident resolution and capacity planning.
Azure Confluent monitoring
Track the performance of Confluent Kafka on Azure and capture alerts such as sustained drops in message counts below certain thresholds.
Business value: Earlier detection of throughput issues in data pipelines that feed services and models.
More container observation capabilities are available to Virtana customers. Read the Container Observability technical blog to learn more about the rest.
Infrastructure Observability: Resilience and topology you can act on
Dell PowerProtect integration
Infrastructure Observability now ingests metrics and inventory from Dell PowerProtect appliances using native interfaces (Agentless). A topology view shows how backup components relate to protected assets.
Business value: Understand backup health in context with the rest of the stack, verify protection status during incidents, and confirm RPO and RTO expectations with concrete data.
Dell ObjectScale integration (Early Access)
Infrastructure Observability now discovers ObjectScale clusters and collects capacity, performance, and health metrics at the cluster level. Futures include tenant levels, including namespace and bucket detail. A topology view maps front-end endpoints, nodes, and dependent services so you can correlate storage behavior with upstream applications.
Business value: Detect saturation and skew before they impact services, tie object storage latency and errors to user-facing symptoms, and plan capacity with evidence rather than guesswork.
A lot more capabilities came to Virtana Infrastructure Observability customers in this release. Read the Infrastructure Observability technical blog to find out more.
Service Observability: Faster triage, clearer reporting
Cisco Meraki Cloud-to-Cloud Connection
High-capacity, lightweight collection for Meraki brings scalable monitoring without on-prem agents, even during maintenance windows.
Business value: Gain dependable visibility for distributed sites and branch networks, helping teams prove or rule out the network during incidents.
Advanced Kubernetes monitoring for Service Observability
Cross-launch from Service Observability into Container Observability to drill into pod-level telemetry, including logs and traces.
Business value: Shorten mean time to resolution by moving from an alert to detailed Kubernetes evidence without changing tools.
The Virtana Service Observability team has been busy delivering even more advancements into the platform, so read the Service Observability technical blog to learn more.
Conclusion
Running AI in production demands evidence. You get cost truth from token tracking, efficiency truth from GPU power, thermals, and utilization, and operational truth from correlated signals across Kubernetes, storage, backups, and network. That gives you faster triage, fewer misdiagnoses, lower run costs, and reports that finance can stand behind.
To see these updates against your AI workloads or discuss deployment options, contact your Virtana sales representative or request a demo.
Virtana Insight
