Cisco 300-640 DCAI Practice Questions with Explanations

Free Cisco 300-640 DCAI practice questions. 14 of them, each with the correct answer, a full explanation, and the reason every other option is wrong.

300-640 practice questions

Question 1

Which result is provided through image recognition using AI?

  • A. anonymization of audit trails for compliance objectives
  • B. detection and labeling of visual elements from digital sources
  • C. scheduling of management tasks during maintenance periods
  • D. routing of multicast messages across network boundaries
Show answer and explanation ▾

Correct answer: B

Image recognition analyzes images or video to detect objects and scenes and label them, such as people or vehicles.

Why the other options are wrong:

  • A. Anonymizing audit trails is a compliance task, not a visual one.
  • C. Scheduling management tasks is a planning function unrelated to images.
  • D. Routing multicast messages is a network function, not an AI image result.

Question 2

How does RAG enhance the capabilities of LLMs?

  • A. by removing the data preparation and cleaning step
  • B. by removing inherent biases from LLM outputs
  • C. by providing access to proprietary data
  • D. by retaining the LLM by altering its core parameters
Show answer and explanation ▾

Correct answer: C

RAG lets a language model retrieve information from outside sources, such as private company data, when it answers. This gives the model access to proprietary data without changing its parameters.

Why the other options are wrong:

  • A. RAG does not remove data preparation; the knowledge sources still need to be prepared.
  • B. RAG does not remove bias from model output.
  • D. RAG works without altering or retraining the model parameters.

Question 3

An organization deploys a new AI training fabric that uses RoCEv2 for GPU communication. The network architect designs the QoS configuration to ensure reliable RDMA transport and must meet these requirements:

Support 256 GPU servers with RDMA connectivity.

Prevent any packet loss that causes RDMA connection failures.

Maintain consistent low-latency communication with a target of less than 10 microseconds.

Use industry-standard protocols and configurations.

Which configuration ensures that RoCEv2 operates as a lossless transport?

  • A. Implement traffic shaping to smooth RDMA traffic bursts and prevent congestion.
  • B. Enable Priority Flow Control on the traffic class carrying RoCEv2 packets.
  • C. Set the RDMA traffic to use the highest DSCP value to ensure priority forwarding.
  • D. Configure weighted fair queuing to prioritize RDMA traffic over other traffic classes.
Show answer and explanation ▾

Correct answer: B

Priority Flow Control pauses the traffic class that carries RoCEv2 when buffers fill, so packets are not dropped. RDMA depends on this lossless behavior to avoid connection failures.

Why the other options are wrong:

  • A. Traffic shaping smooths bursts but does not guarantee that no packets are dropped.
  • C. A high DSCP value marks priority but does not make the transport lossless.
  • D. Weighted fair queuing shares bandwidth between classes and does not prevent packet loss.

Question 4

Which set of statements describes Quantized Congestion Notification?

  • A. Priority Flow Control reacts first to mitigate congestion, and Explicit Congestion Notification acts as a fail-safe to prevent traffic drops if Priority Flow Control is insufficient. A congestion notification packet sent toward the source by a network switch in response to receiving Explicit Congestion Notification with congestion experienced bits set.
  • B. Explicit Congestion Notification reacts first to mitigate congestion, and Priority Flow Control acts as a fail-safe to prevent traffic drops if Explicit Congestion Notification is insufficient. A congestion notification packet sent toward the source by hosts and the network switch in response to receiving Explicit Congestion Notification with congestion experienced bits set.
  • C. Explicit Congestion Notification reacts first to mitigate congestion, and Priority Flow Control acts as a fail-safe to prevent traffic drops if Explicit Congestion Notification is insufficient. A congestion notification packet sent toward the source by a host in response to receiving Explicit Congestion Notification with congestion experienced bits set.
  • D. Priority Flow Control reacts first to mitigate congestion, and Explicit Congestion Notification acts as a fail-safe to prevent traffic drops if Priority Flow Control is insufficient. A congestion notification packet sent toward the source by hosts and the network switch in response to receiving Explicit Congestion Notification with congestion experienced bits set.
Show answer and explanation ▾

Correct answer: C

In the QCN-style flow used with RoCEv2, ECN reacts first and Priority Flow Control is the fail-safe. The receiving host sends the congestion notification packet back toward the source when it sees packets with the congestion experienced bits set.

Why the other options are wrong:

  • A. It reverses the roles, since ECN acts first and PFC is the fail-safe.
  • B. The notification packet is sent by the host, not by hosts and the network switch.
  • D. It reverses the roles of ECN and PFC and also names the switch as a sender of the notification packet.

Question 5

A network engineer analyzes RoCEv2 traffic flows in an AI training cluster and notices that all RoCEv2 packets have the Don’t Fragment bit set in the IP header.

Why must RoCEv2 set the DF bit in all packets?

  • A. to enable Path MTU Discovery so that the optimal packet size can be negotiated
  • B. to indicate to intermediate switches that the packet contains RDMA traffic requiring special handling
  • C. to ensure that packets always take the same path through ECMP routing for consistent latency
  • D. to prevent IP fragmentation that would break the InfiniBand sequence numbering and ordering guarantees
Show answer and explanation ▾

Correct answer: D

RoCEv2 carries InfiniBand transport semantics, which depend on packet sequence numbering and ordering. Fragmentation would break those guarantees, so the DF bit is set on all packets.

Why the other options are wrong:

  • A. RoCEv2 sets DF to prevent fragmentation, not to negotiate packet size with Path MTU Discovery.
  • B. The DF bit does not signal RDMA traffic to switches.
  • C. The DF bit does not control ECMP path selection.

Question 6

Which components are part of the HyperFabric AI solution?

  • A. Cisco UCS servers combined with high-speed networking switches and third-party storage
  • B. integrated compute and networking hardware optimized for AI workloads, with optional cloud management
  • C. Cisco UCS servers with NVIDIA GPUs, switches, optional VAST Data Storage and cloud-managed networking
  • D. combination of Cisco networking hardware and software-defined storage designed for AI scalability
Show answer and explanation ▾

Correct answer: C

HyperFabric AI combines Cisco UCS servers with NVIDIA GPUs, switches, optional VAST Data storage and cloud-managed networking into one AI infrastructure solution.

Why the other options are wrong:

  • A. It leaves out the NVIDIA GPUs and the VAST Data storage and cloud management that make up the solution.
  • B. It is too generic and does not name the GPU servers, switches and storage in the solution.
  • D. It describes software-defined storage, which is not how the solution is composed.

Question 7

A data center administrator for a medium-sized manufacturing facility uses Intersight to perform unattended operating system installation on already discovered X-Series servers with server profiles assigned.

Which set of prerequisites must be met for installing an operating system on a server in Cisco Mode?

  • A. configuring a firmware policy adding an operating system configuration file to the software repository
  • B. adding an OS image to the software repository adding an operating system configuration file to the software repository
  • C. adding an OS image to the software repository adding a server configuration utility to the software repository
  • D. configuring a virtual KVM policy adding a server configuration utility to the software repository
Show answer and explanation ▾

Correct answer: B

An unattended installation from Intersight needs the operating system ISO file in the software repository and an operating system configuration file in the repository that supplies the answers for the unattended setup.

Why the other options are wrong:

  • A. A firmware policy controls server firmware, and this set leaves out the operating system ISO file.
  • C. This set leaves out the operating system configuration file that drives the unattended setup.
  • D. A virtual KVM policy controls remote console access, and this set leaves out the operating system ISO file and configuration file.

Question 8

A national law firm is deploying a RAG solution for their legal research team to more efficiently use their 50 TB of confidential data and documents when preparing new contracts and briefs. They decided on a UCS X-Series chassis equipped with four GPUs and six X210c servers.

Which storage solution fits this deployment?

  • A. IP-based storage array
  • B. local storage in a JBOD configuration
  • C. tape storage
  • D. local storage in a RAID 5 configuration
Show answer and explanation ▾

Correct answer: A

A 50 TB shared data set serving six servers calls for external shared storage. An IP-based storage array gives all the X210c servers common access to the documents, which local drives cannot do.

Why the other options are wrong:

  • B. Local JBOD storage is not shared across servers and offers no redundancy for 50 TB of confidential data.
  • C. Tape storage is far too slow for interactive retrieval in a RAG workload.
  • D. Local RAID 5 storage is not shared between the six servers and does not suit this data volume.

Question 9

An engineer must optimize the performance of an AI inference workload running on a Cisco UCS C-Series rack server. The workload experiences intermittent latency spikes, and Intersight system logs show frequent thermal event entries.

Which troubleshooting action must be taken first to address the performance issues on the Cisco UCS server?

  • A. Review the resource utilization of the AI workload (CPU, memory, and GPU) to identify potential bottlenecks and optimize resource allocation.
  • B. Examine the storage configuration, including the RAID configuration and disk health, to ensure that the storage subsystem is not causing latency issues.
  • C. Check the power supply status and power consumption of the UCS server using UCS Manager to ensure that the server has sufficient power.
  • D. Inspect the physical environment of the server, including ambient temperature and airflow, and replace any failed fans.
Show answer and explanation ▾

Correct answer: D

Thermal event entries point to a cooling problem, so the first step is to inspect the physical environment, including ambient temperature and airflow, and replace any failed fans. Overheating causes throttling that shows up as latency spikes.

Why the other options are wrong:

  • A. Reviewing resource utilization looks at the workload, not the thermal events in the logs.
  • B. Storage configuration does not explain thermal event entries.
  • C. Power supply status is not what the thermal event entries point to.

Question 10

An engineer made changes to the corporate DNS, which caused a Cisco UCS X-Series domain to lose connectivity to Intersight. To restore connectivity, the management interface DNS and NTP settings must be updated using the device console UI.

Which configuration change must be made in Intersight to allow the updated DNS configuration to persist?

  • A. Edit the IMC access policy associated to the fabric interconnects.
  • B. Create a device connector policy and apply it to the blade chassis.
  • C. Create a network connectivity policy and apply it to the blade chassis.
  • D. Edit the domain profile deployed to the fabric interconnect.
Show answer and explanation ▾

Correct answer: D

The management interface DNS and NTP settings for a UCS X-Series domain are defined in the domain profile deployed to the fabric interconnects. Editing that profile makes the console change persist.

Why the other options are wrong:

  • A. An IMC access policy covers server management access, not the domain's DNS settings.
  • B. A device connector policy is not applied to the blade chassis to hold domain DNS settings.
  • C. A network connectivity policy applied to the chassis does not hold the fabric interconnect DNS settings.

Question 11

What describes inference traffic patterns in AI deployments?

  • A. optimizes data storage solutions for data lakes
  • B. simplifies complex neural network architectures
  • C. provides real-time decision making
  • D. eliminates the need for data preparation
Show answer and explanation ▾

Correct answer: C

Inference means a trained model receives live input and returns a prediction with low delay, which is what lets applications make decisions in real time.

Why the other options are wrong:

  • A. Optimizing data storage for data lakes is a storage design task, not what inference traffic does.
  • B. Inference runs a model that already exists and does not simplify neural network architecture.
  • D. Data still has to be prepared and cleaned, so inference does not remove that need.

Question 12

What is the process involved in workload scheduling for AI environments?

  • A. assignment of tasks to compute elements for optimized performance
  • B. optimization of licensing structures for compliance
  • C. aggregation of device statistics for reporting
  • D. generation of backup images for restoration operations
Show answer and explanation ▾

Correct answer: A

Workload scheduling assigns AI jobs to available compute such as GPUs, CPUs and nodes, taking resources and job needs into account to improve performance.

Why the other options are wrong:

  • B. Licensing optimization is a procurement and compliance matter, not scheduling.
  • C. Aggregating device statistics is monitoring and reporting.
  • D. Generating backup images is data protection, not scheduling workloads.

Question 13

What does workload distribution offer in an AI infrastructure with local and external resources?

  • A. isolation of analytic functions within remote processing clusters
  • B. distribution of computational tasks across multiple locations
  • C. restriction of automation to local storage and performance measures
  • D. enforcement of fixed security policies on distribution endpoints
Show answer and explanation ▾

Correct answer: B

Workload distribution spreads computational tasks across multiple locations, both local and external, so AI workloads can use the available resources.

Why the other options are wrong:

  • A. Isolating analytic functions in remote clusters describes separation, not distribution of tasks.
  • C. Workload distribution is not restricted to local storage and performance measures.
  • D. Enforcing fixed security policies is a security matter, not what workload distribution offers.

Question 14

A medium-sized breadmaking operation must deploy a fleet of GPU servers that meet these requirements for their flagship application: policy managed through cloud-based interface low latency between specific pairs of GPUs at least 2 GPUs per server all

GPUs should be same modelIntel CPU -well suited for Inferencingcustomer is power consciousWhich server meets the requirements?

  • A. C245 M8
  • B. X210C M8 with x580 PCIe node
  • C. C845A M8
  • D. C885A M8
Show answer and explanation ▾

Correct answer: B

The X210c M8 compute node with an X580 PCIe node uses Intel CPUs, supports multiple identical GPUs per server, and is managed through the cloud-based Intersight interface.

Why the other options are wrong:

  • A. The C245 M8 is an AMD processor server, so it does not meet the Intel CPU requirement.
  • C. The C845A M8 is an AMD processor server, so it does not meet the Intel CPU requirement.
  • D. The C885A M8 is an AMD processor server, so it does not meet the Intel CPU requirement.

That was 14 of 60.

The full Cisco 300-640 DCAI pack has all 60 questions, each with the answer, the explanation and why the other options are wrong, plus a questions-only copy for timed runs. $39.00 USD, paid once, with free monthly updates and a pass-or-your-money-back guarantee.

Get the full pack

Not ready yet? Get the first 16 free as one PDF by email. Get the Half Pack →

About the 300-640 exam

These 14 Cisco 300-640 DCAI questions are free here. The free PDF has 16, free by email.

They are drawn from the same bank as the full 300-640 pack, which has 60 questions in total.

Back to blog