Announcing Cisco Nexus Dashboard 4.3: Nexus One Architecture, AI Observability and Operational Efficiency

Consolidation, and now efficiency

Data center networking teams are being asked to operate more fabrics and distinct architectures all in parallel. A single organization may run ACI in one facility, VXLAN EVPN in another.  Additional data centers may house dedicated AI fabrics for both training and inference clusters alongside SAN and media environments. Each of those has historically carried its own tooling, its own telemetry and its own operational procedures.

Cisco Nexus Dashboard 4.3 continues the consolidation work of the previous few releases, with an emphasis in this release falling into four areas. First, streamlined AI operations including the monitoring of AI workloads and leveraging AI to operate the network. Second, this release treats VXLAN and ACI as a common operational domain in support of the Nexus One initiative. Third, security and network teams are provided a shared view of device exposure and protection status. Fourth, this release removes steps from routine work that has historically required more effort than it should.

What's new in version 4.3?

This release puts an emphasis on operational efficiency, increasing ease of use, shortening common workflows and giving operators more of what they need in fewer places.

Unified operations across VXLAN and ACI

The VXLAN and ACI fabric group experience is meaningfully expanded in 4.3 as the focal point of Cisco's Nexus One architecture. This allows operators to manage distinct VXLAN-EVPN and ACI domains through a single operational interface. The platform now enables automated EPG to ESG migration for ACI, global Search and Explore functionality to query endpoints, policy definitions and operational states across connected fabric types. Teams can also leverage the Connectivity Analysis and Traffic Analytics tools for troubleshooting and traffic visibility across distinct domains while still monitoring anomalies, advisories and inter-fabric connectivity health across all member fabrics.

This is the clearest expression thus far of operating two previously separate domains as a single cohesive environment. It does not merge the architectures; it is not intended to. It does, however, simplify operations regardless of which fabric type an endpoint happens to live in.

End-to-end AI job monitoring

AI job monitoring in Nexus Dashboard now reaches further into the server, expanding the AI job visibility features introduced in the previous release. In addition to GPU and switch interface data, 4.3 collects server NIC telemetry, including CRC errors, discards, bandwidth, NACKs, CNPs, operational state and link speed.

The practical value of these enhancements is in determining whether a runtime problem is rooted in the server or in the network fabric. Correlating these metrics for the same job can remove a lot of guesswork from an investigation by removing the need to reconcile timestamps and monitoring stats across a wide variety of tools while a training run is degraded, ultimately reaching resolution faster.

AI-ready operations through MCP

Nexus Dashboard 4.3 introduces a Model Context Protocol, or MCP, server natively integrated into the platform. This establishes a secure interface connecting Nexus Dashboard to Cisco Cloud Control and/or a customer's own large language models, or LLMs. By utilizing this MCP integration, organizations can implement AI-assisted workflows and retrieve network telemetry locally. This setup ensures that enterprise data and custom AI models remain strictly within the customer's operationally secure boundaries.

The architectural distinction here is worth drawing out. Most currently available AI-assisted operations require infrastructure telemetry to leave the environment to reach a hosted model. Nexus Dashboard's native MCP server inverts that concept by bringing the model to the data. Organizations that cannot send telemetry externally now have a viable path to true agentic operations.

WWT is currently testing the use of LLM queries against Nexus Dashboard 4.3 in the Advanced Technology Center. We expect the immediate value to be in 'read and query' workflows rather than in active change automation; that capability will come later.

A consolidated view of device security

Live Protect arrived in Nexus Dashboard 4.2, leveraging eBPF to apply compensating controls to NX-OS without a switch reload or a maintenance window. Version 4.3 adds a new reporting layer around the process. The new Overview tab on the Device Security page brings together fabric exposure data and active protection status. This consolidated pane provides visibility into Cisco security advisories alongside Live Protect deployment states across supported switches in active inventory. Security policy hit counts and historical trends are presented together to help teams prioritize remediation efforts.

The Device Security section of Nexus Dashboard is important to the full breadth of Network and Security teams within an organization. It provides stakeholders with a way to analyze fleet exposures, confirm that mitigations were deployed and determine if those compensating controls have been triggered by malicious activity.

Software operations in monitor mode

Software image management, or SWIM, can now be used for fabrics in monitor mode. While this may appear as a small change in behavior, the value is significant. Teams evaluating Nexus Dashboard, or running production fabrics in read-only mode, can now stage and organize OS upgrades without making a change to the administrative state of the fabrics while preparing for their maintenance windows.

Smarter anomaly management

The sheer volume of alerts in any telemetry-rich platform is a persistent frustration amongst operators. Nexus Dashboard 4.3 takes two steps to rectify this. First, users can simply suppress anomalies that are not relevant to their environment.  Users can also create rules directly from anomalies rather than building them separately and then apply these suppression rules globally or across specific fabrics. 

Second, is OS upgrade-aware event correlation. During maintenance windows, Nexus Dashboard's correlation engine automatically attributes downstream telemetry events to the root upgrade event, preventing unnecessary notification storms. Anyone who has watched an anomaly count spike during a planned upgrade window will recognize the value of identifying and collapsing that noise based on the actual cause.

New features quick hits

• Connected TAC is expanded across LAN and SAN, no longer requiring active software telemetry, and is managed centrally from System Settings.
• Nexus Dashboard can onboard the Cisco Hypershield controller to automate security policy enforcement, zone segmentation and traffic steering on Cisco's DPU-enabled Smart Switches.
• SONiC OS support extended to Cisco N9164E-NS4-O and N93108TC-FX3
• Additional hardware support: Cisco N9348Y12C-SE1, N9164E-NS4-O, N9K-C9804 and N9K-C9808, Catalyst C9300L-24P-4G and C9300L-48P-4G and Cisco 8455-G2 and 8475-G2 secure routers

Operational considerations and behavior changes

• The Cisco Nexus 9000 Series is rebranded as the Cisco N9000 Series, with N9200, N9300, N9400, N9500, N9800 and N9300 Smart Switches accordingly.
• Several previously supported virtual cluster types are no longer supported as greenfield deployments. However, they remain supported as upgrade targets. The 5-node virtual cluster with 500G app nodes will not be supported as an upgrade option in the next release, presumably 4.4. If you're running this deployment model, start migration planning now; WWT can help scope that process.
• Traffic Analytics compatibility mode now requires Network Time Protocol (NTP) configuration across target switches instead of Precision Time Protocol (PTP).
• All inbound API calls are validated against published OpenAPI schemas prior to backend execution, and non-compliant payloads accepted in previous releases will now be rejected.

Learn more

For a broader look at the platform and how we run it at WWT, see Inside WWT's ATC: Operating at Scale with Cisco Nexus Dashboard where we cover what Nexus Dashboard is, where it fits and how the Advanced Technology Center operates it day to day. For a step-by-step walkthrough of initial cluster setup and configuration, see Deploying a Nexus Dashboard Cluster: A Guided Walkthrough or our hands-on lab to build VXLAN fabrics managed by Nexus Dashboard.

How WWT helps

World Wide Technology (WWT) is the premier resource for helping customers design and build their modern data centers. With decades of experience in data center networking and modern security initiatives, WWT has a unique combination of expertise, partner relationships and hands-on learning opportunities. The Labs and Learning Paths on our digital platform can help you gain tangible experience with these technologies, while WWT Research can further provide real-world use cases and white papers from customers we have helped along their IT transformation journeys. If you are interested in a more customized lab environment, our world-renowned technology proving ground, the Advanced Technology Center, is the place to do it.

Contact your WWT Account Team or send us a note to learn more about how WWT can support your organization on its infrastructure modernization journey.