Unlocking AI Performance: Splunk Observability for Cisco Secure AI Factory with NVIDIA

[ad_1] Synthetic intelligence is reshaping each trade, and unlocking its full potential requires infrastructure that's sturdy, scalable, safe, and observable. As organizations develop their AI initiatives, managing complicated workloads and guaranteeing constant efficiency develop into mission-critical.

That is the place Cisco AI PODs, the foundational constructing blocks of Cisco Safe AI Manufacturing facility with NVIDIA, mixed with the deep visibility of Splunk Observability Cloud, ship a robust resolution for constructing and operating fashionable AI environments.

Cisco AI PODs: The muse for AI innovation Cisco AI PODs are modular, versatile, and scalable AI infrastructure designed to speed up time to worth for AI initiatives. They permit organizations to deploy production-grade AI environments shortly—however to maintain these environments operating optimally, groups want complete perception into efficiency and well being.

How will you detect points early, troubleshoot effectively, and deal with delivering enterprise outcomes as a substitute of spending time addressing pressing manufacturing points? That’s the place observability turns into indispensable.

Splunk Observability: Your eyes and ears inside AI PODs <img loading="lazy" decoding="async" class="lazy lazy-hidden aligncenter size-medium_large wp-image-480514" data-lazy-type="image" src="https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-768x532.png" alt="Slide for Cisco Secure AI Factory with NVDIA - displaying a vertically integrated deployment option built on Cisco AI PODs and including NVIDIA AI software, Kubernetes platform, Cisco AI networking, Cisco Compute with NVIDIA Accelerated Computing and partner storage. Additional pillars include Cisco Security and Splunk Observability." width="768" height="532" srcset="" sizes="auto, (max-width: 768px) 100vw, 768px"/><img loading="lazy" decoding="async" class="aligncenter size-medium_large wp-image-480514" src="https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-768x532.png" alt="Slide for Cisco Secure AI Factory with NVDIA - displaying a vertically integrated deployment option built on Cisco AI PODs and including NVIDIA AI software, Kubernetes platform, Cisco AI networking, Cisco Compute with NVIDIA Accelerated Computing and partner storage. Additional pillars include Cisco Security and Splunk Observability." width="768" height="532" srcset="https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-300x208.png 300w, https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-1024x709.png 1024w, https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-768x532.png 768w, https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-1536x1064.png 1536w, https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min-770x533.png 770w, https://blogs.cisco.com/gcs/ciscoblogs/1/2025/10/GTCxNvidia_1_figure-1-min.png 2000w" sizes="auto, (max-width: 768px) 100vw, 768px"/>

Splunk Observability Cloud delivers end-to-end visibility throughout each layer of Cisco AI PODs—from bodily infrastructure to Kubernetes to the AI purposes layer.

It’s not nearly knowledge assortment. Splunk turns metrics, traces, and logs into actionable insights, serving to groups detect, troubleshoot, and resolve points in seconds.

We’re excited to introduce a brand new Splunk Dashboard purpose-built for observability throughout your complete AI POD stack.

What the brand new Splunk Dashboard brings to Cisco AI PODs

Unified Kubernetes cluster monitoring – Get a single view of all Kubernetes clusters, together with Pink Hat OpenShift operating on AI PODs. Deep host-level insights – Monitor the efficiency of particular person Cisco UCS servers, together with CPU, reminiscence, disk, and community utilization. AI POD infrastructure dashboard – Observe vital metrics like GPU utilization, GPU reminiscence utilization, energy, and community efficiency, integrating knowledge from Cisco Intersight and Cisco Nexus. Streaming analytics benefit – Leverage Splunk’s real-time streaming analytics to attain quicker detection and near-instant “time to glass.”

Whereas Cisco AI PODs present modular, scalable infrastructure for enterprise AI, every AI POD will also be monitored individually. This enables groups to achieve detailed perception into the distinctive efficiency metrics and workloads of a selected deployment. Listed below are some screens from the Splunk Dashboard for AI PODs to assist visualize the monitoring capabilities. By aggregating the variety of enter and output tokens processed by the big language mannequin (LLM) operating on an AI POD, Splunk is ready to calculate an approximate price for token utilization over time:

Splunk additionally pulls in metrics from Cisco Intersight, to supply visibility to lively alarms associated to the monitored AI POD, and key UCS metrics comparable to UCS host energy, temperature, and fan velocity:

The Nexus dashboard gives perception into the interfaces configured on every Nexus change, the transmit errors and drops, and the information transferred between storage and compute nodes:

An actual-world state of affairs: Diagnosing LLM latency Think about an software operating on a Cisco AI POD using an LLM for person queries. Instantly, response instances on the appliance spike. Right here’s how Splunk Observability Cloud helps resolve it in minutes:

Alert triggered – Splunk detects excessive response instances and raises an alert. Hint evaluation – The service map highlights that the majority latency happens inside /v1/chat/completions calls to the LLM. Infrastructure view – The AI POD dashboard reveals that solely one of many 4 obtainable GPUs is lively and totally utilized. Actionable perception – You reconfigure the workload to make use of all GPUs—immediately restoring efficiency.

The NVIDIA connection: Powering clever workloads Splunk Observability additionally screens key NVIDIA AI Enterprise parts—together with the NVIDIA NIM operator and NVIDIA NIMs microservices for LLM inferencing—guaranteeing the NVIDIA software program stack performs at its finest.

FedRAMP and authorities readiness: Splunk’s present path in the direction of reaching FedRAMP Average for Splunk Observability Splunk stays a trusted associate in authorities digital transformation, empowering companies to ship safe, resilient, and clever providers via cloud and customer-managed options. Constructing on the success of Splunk Cloud Platform—licensed at FedRAMP Excessive and DoD Influence Stage 5, and listed on the StateRAMP (dba GovRAMP) Approved Merchandise Listing—Splunk continues to spend money on increasing our FedRAMP program to fulfill evolving public sector wants. As beforehand introduced, Splunk Observability Cloud has already obtained “In Course of” designation and awaits full authorization to function on the Average stage from the FedRAMP Program Administration Workplace. Splunk stays dedicated to supporting the safety and mission success of all our authorities prospects.

Observability: A cornerstone of Cisco Safe AI Manufacturing facility with NVIDIA In Cisco Safe AI Manufacturing facility with NVIDIA, observability is just not elective—it’s foundational.

By delivering deep, real-time insights throughout infrastructure and purposes, Splunk Observability Cloud enhances:

Operational effectivity Useful resource optimization Reliability and uptime Safety posture

This holistic visibility is crucial for constructing, working, and securing complicated AI pipelines at scale.

Conclusion Cisco AI PODs ship the sturdy, scalable infrastructure required for as we speak’s demanding AI workloads. When paired with Splunk Observability Cloud, organizations achieve unmatched visibility and management—enabling fast troubleshooting, optimum efficiency, and quicker innovation.

Splunk Observability types a core pillar of Cisco Safe AI Manufacturing facility with NVIDIA, empowering companies to construct and run AI with confidence, velocity, and safety.