Automation and Artificial Intelligence

A data center team is building a dedicated back-end fabric to interconnect 256 GPUs across multiple servers for large-model training. The AI workload is dominated by all-reduce collective operations that generate synchronized, bursty east-west traffic where the slowest flow dictates job completion time. Which fabric design characteristic is MOST important to prioritize to minimize tail latency and avoid degrading GPU utilization?

A data center team is designing an AI training cluster with servers containing 8 GPUs each. Within a single server, GPUs must exchange gradients at maximum bandwidth with the lowest possible latency, bypassing CPU and PCIe bottlenecks. Across servers, GPUs connect through a lossless Ethernet fabric using RoCEv2. Which technology provides the intra-server (scale-up) high-speed GPU-to-GPU interconnect that the design should specify?

A data center automation engineer maintains Cisco Nexus 9000 switches using Ansible with the cisco.nxos collection. Before applying a large VLAN and interface configuration change to production leaf switches, the engineer wants to preview exactly which tasks would report a change WITHOUT altering any device state, and also wants confidence that re-running the playbook after a successful apply will report no further changes. Which combination of Ansible characteristics satisfies both requirements?

A network automation engineer is storing an Ansible playbook and its variable files in a shared Git repository used to configure Cisco Nexus 9000 switches. The variable files contain BGP neighbor authentication keys and device login credentials in plaintext. Security policy requires that these secrets not be readable in the repository while still allowing the playbook to run non-interactively in a CI/CD pipeline. Which approach meets the requirement?

A data center engineer is developing a RESTCONF client to manage interface descriptions on a Nexus 9000 switch running NX-OS. The client currently sends an HTTP PUT to the resource URI for a specific interface containing only the 'description' leaf. After the call, the engineer notices that other previously configured attributes on that same interface node (such as MTU and speed settings within the same YANG container) have been reset to their defaults. What is the most likely cause, and how should the request be changed to update only the description without affecting other attributes?

A network engineer wants to automate configuration of a Cisco Nexus 9000 switch from a Python script running on an external server. The requirement is to send existing CLI commands (e.g., 'show version', 'configure interface') and receive structured JSON responses over an HTTPS session, without having to rewrite the logic around a YANG data model. Which on-switch interface best meets this requirement?

A network automation engineer is writing an Ansible playbook to configure a fleet of Nexus 9000 switches. The team wants the playbook to interact with the switches over an HTTPS-based programmatic interface rather than opening interactive SSH CLI sessions, and they have already enabled the 'nxapi' feature on all devices. In the playbook's connection settings, which value must be assigned to the 'ansible_connection' variable so tasks are delivered through the NX-API back end?

A network engineer must build an automated inventory audit that collects the running software version, serial number, and installed module list from 40 Nexus 9000 switches, then generates a report. The engineer wants to use Ansible without writing custom parsing logic or issuing raw show commands that require regex screen-scraping. Which approach best meets these requirements?

A data center engineer needs to install a custom third-party monitoring agent (packaged as an RPM) directly on the Nexus 9000 switch's underlying Linux host so it runs in the native NX-OS environment rather than in an isolated container. Which on-box automation capability should the engineer enable and use to accomplish this?

A network automation engineer wants to programmatically retrieve the full operational and configuration state from a Nexus 9000 switch and integrate it into a Python-based orchestration workflow. The workflow must parse structured data, validate it against a YANG data model, and support model-driven configuration changes over a single standardized transport. Which approach best satisfies these requirements?

A network engineer wants a Nexus 9000 switch to automatically shut down interface Ethernet1/10 and generate a custom syslog message whenever that interface flaps more than three times within 60 seconds, without any external server involvement. Which on-box automation approach directly satisfies this requirement?

A network engineer writes an NX-OS EEM applet to disable a specific interface during scheduled maintenance. The applet must be triggered manually by an operator on demand rather than by any system event, and it must be runnable from the CLI with 'event manager run'. Which configuration element is required for this applet to be manually executable?

A network engineer wants an NX-OS device to automatically collect a detailed diagnostic bundle whenever an interface flaps more than five times in 60 seconds. The collection logic is complex, requiring conditional parsing of multiple show commands and formatting output into a JSON file stored in bootflash. The engineer plans to use an EEM policy to trigger this. Which approach best satisfies the requirement while keeping the trigger logic maintainable?

A network engineer wants to run a persistent, Linux-based Python 3 environment directly on a Nexus 9000 switch to execute custom monitoring scripts that use standard PyPI libraries installed via pip. The script must survive supervisor switchovers and run in an isolated container without affecting NX-OS. Which on-box automation environment satisfies these requirements?

A network engineer is developing a Python script to configure multiple Nexus 9000 switches through the NX-API REST interface. The script must authenticate once and reuse the session for subsequent configuration calls to avoid re-sending credentials with every request. Which mechanism should the engineer use to maintain the authenticated session across multiple NX-API REST calls?

A data center engineer is deploying streaming telemetry on Cisco Nexus 9000 switches to feed a centralized collector. The security team requires that the switches never listen for inbound connection requests on any telemetry port, and the collector must reside behind a firewall that only permits outbound-initiated sessions from the switches. The engineer must also ensure the transport survives collector restarts by having the switch re-establish the session automatically. Which telemetry configuration approach meets these requirements?

A network engineer is deploying model-driven telemetry on a Nexus 9000 switch to stream interface counters to an external collector. The requirement is that the switch itself initiates the session to the collector's IP and TCP port, and the data must be serialized in a compact binary format for efficient parsing at high sample rates. Which combination of telemetry transport mode and encoding should the engineer configure?

A data center engineer must run a custom Python script directly on a Nexus 9000 switch that parses interface counters, needs its own isolated Linux environment with pip-installed libraries (such as requests), and must be able to reach the management network to POST results to an external REST collector. The solution must not require an external server and must keep the script's runtime dependencies separate from the NX-OS system software. Which on-box capability best meets these requirements?

A network engineer must deploy identical interface descriptions and VLAN assignments across 40 Nexus 9000 switches, where each switch has different values (hostname, management IP, uplink descriptions) pulled from a YAML data file. The engineer wants a maintainable workflow that separates the static configuration structure from the per-device variable data, then renders complete device configurations for each switch. Which approach best meets these requirements?

A network engineer writes an on-box Python script to run on a Nexus 9000 switch. The script must execute the show CLI command 'show interface brief', parse the returned data programmatically, and iterate over each interface. The engineer wants the command output returned as structured data rather than raw text so the parsing logic stays simple. Which approach within the on-box Python interpreter meets this requirement?