cert
cert-1
cert-2

Pass Cisco 300-640 Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
300-640 Exam - Verified By Experts
300-640 Premium File

300-640 Premium File

$89.99
$98.99
  • Premium File 60 Questions & Answers. Last Update: Nov 10, 2026

Whats Included:

  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates
 
$98.99
$89.99
accept 10 downloads in the last 7 days
block-screenshots
Cisco 300-640 Practice Test 34

Question 330

The acceptance board must decide whether an AI-infrastructure response to enhanced transmission selection shares is fit for controlled operation. The fictional operating facts are: An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. Signoff cannot rely on a good-looking utilization graph, a change ticket alone or an unmeasured promise. The response must include an appropriate governing engineering principle, a scoped intervention and a separate measurable acceptance criterion. Choose EXACTLY THREE statements that meet those roles.

  1. Maximum approved training load stays within electrical limits without involuntary power capping or host resets.
  2. ECMP spreads eligible flows across paths but a few very large flows can collide; validate hashing and path capacity before changing policy.
  3. Each class achieves its minimum agreed service during simultaneous representative demand.
  4. Apply a supported UCS power policy after confirming rack and chassis electrical budgets.
  5. ETS partitions link bandwidth among configured traffic classes; it does not guarantee arbitrary per-flow latency.
  6. Configure supported ETS class weights to protect essential shares without starving management and storage traffic.

Explanation

Correct Answer

C:

An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. Each class achieves its minimum agreed service during simultaneous representative demand. Configure supported ETS class weights to protect essential shares without starving management and storage traffic.

E:

An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. ETS partitions link bandwidth among configured traffic classes; it does not guarantee arbitrary per-flow latency. Configure supported ETS class weights to protect essential shares without starving management and storage traffic.

F:

An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. Configure supported ETS class weights to protect essential shares without starving management and storage traffic. ETS partitions link bandwidth among configured traffic classes; it does not guarantee arbitrary per-flow latency.

Incorrect Answers

A:

Maximum approved training load stays within electrical limits without involuntary power capping or host resets. It would not independently demonstrate that the control works: An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. Each class achieves its minimum agreed service during simultaneous representative demand. Configure supported ETS class weights to protect essential shares without starving management and storage traffic.

B:

ECMP spreads eligible flows across paths but a few very large flows can collide; validate hashing and path capacity before changing policy. It is not the rule that supports the proposed change and acceptance test: An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. ETS partitions link bandwidth among configured traffic classes; it does not guarantee arbitrary per-flow latency. Configure supported ETS class weights to protect essential shares without starving management and storage traffic.

D:

Apply a supported UCS power policy after confirming rack and chassis electrical budgets. It does not implement the governing rule for this approval: An inference cluster protects interactive packets but accidentally assigns background data movers all remaining class bandwidth. Configure supported ETS class weights to protect essential shares without starving management and storage traffic. Each class achieves its minimum agreed service during simultaneous representative demand.

Question 331

A data-center AI architecture committee considers rocev2 routing prerequisites in its AI Infrastructure Deployment and Data Management review. At the affected site: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. The committee is deliberately not selecting a vendor product or applying a live change yet. It needs the foundational reasoning that makes the next design decision defensible. Which ONE engineering principle directly governs this situation? Choose ONE answer.

  1. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.
  2. Intersight-driven compute management depends on authorized targets, compatible profiles and predictable policy inheritance.
  3. Server boot and storage policies must match approved boot targets and multipath recovery requirements.
  4. UCS power policy must reflect chassis limits, redundant feed capacity and peak AI workload characteristics.

Explanation

Correct Answer

A:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Incorrect Answers

B:

Intersight-driven compute management depends on authorized targets, compatible profiles and predictable policy inheritance. It does not govern the condition described: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

C:

Server boot and storage policies must match approved boot targets and multipath recovery requirements. It does not govern the condition described: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

D:

UCS power policy must reflect chassis limits, redundant feed capacity and peak AI workload characteristics. It does not govern the condition described: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

Question 332

An infrastructure diagnostician is investigating rocev2 routing prerequisites for the AI Infrastructure Deployment and Data Management team. The reported incident reads: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Competing teams attribute the symptoms to different layers of the AI stack. Before accepting a hypothesis or touching production settings, the investigator must choose a discriminating set of measurements. Which ONE evidence package is MOST relevant? Choose ONE answer.

  1. Review power-cap settings, power telemetry, thermal alarms and PSU redundancy against expected GPU draw.
  2. Correlate ECN marking, switch queue depth, RTT, endpoint congestion control and packet loss during collective bursts.
  3. Compare clock offset, NTP source reachability, drift, log timestamps and monitoring collector time bases.
  4. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Explanation

Correct Answer

D:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

Incorrect Answers

A:

Review power-cap settings, power telemetry, thermal alarms and PSU redundancy against expected GPU draw. That check would not establish the decisive facts for this condition: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

B:

Correlate ECN marking, switch queue depth, RTT, endpoint congestion control and packet loss during collective bursts. That check would not establish the decisive facts for this condition: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

C:

Compare clock offset, NTP source reachability, drift, log timestamps and monitoring collector time bases. That check would not establish the decisive facts for this condition: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Question 333

During a bounded AI-platform change-control review concerning rocev2 routing prerequisites, the site owner reports: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. The proposed work must have a narrow implementation scope, first be validated against the failure evidence and avoid unrelated system-wide tuning. Which ONE intervention should the engineers pilot before any wider rollout? Choose ONE answer.

  1. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.
  2. Choose supported load-distribution behavior and adjust flow placement or links without creating packet reordering hazards.
  3. Apply a staged Intersight policy template to a bounded server group and confirm health before scale-out.
  4. Apply version-supported ECN thresholds for the designated traffic class and validate endpoint response under controlled load.

Explanation

Correct Answer

A:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Incorrect Answers

B:

Choose supported load-distribution behavior and adjust flow placement or links without creating packet reordering hazards. That change would leave the reported problem unresolved: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

C:

Apply a staged Intersight policy template to a bounded server group and confirm health before scale-out. That change would leave the reported problem unresolved: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

D:

Apply version-supported ECN thresholds for the designated traffic class and validate endpoint response under controlled load. That change would leave the reported problem unresolved: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

Question 334

Two independent technical reviewers examine a AI Infrastructure Deployment and Data Management problem involving rocev2 routing prerequisites. Their shared incident log states: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. One reviewer requires measurements capable of separating the observed problem from neighboring failure modes; the other requires a bounded action that follows those measurements. Choose EXACTLY TWO statements that together supply those two parts.

  1. Compare UCS system class, vNIC QoS policy, host packet marking and leaf switch queue counters.
  2. Correlate ECN marking, switch queue depth, RTT, endpoint congestion control and packet loss during collective bursts.
  3. Define trusted classification boundaries and validate consistent class-to-queue behavior across endpoints and switches.
  4. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.
  5. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

Explanation

Correct Answer

D:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

E:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

Incorrect Answers

A:

Compare UCS system class, vNIC QoS policy, host packet marking and leaf switch queue counters. It would not provide the evidence needed for the paired diagnosis and follow-up: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

B:

Correlate ECN marking, switch queue depth, RTT, endpoint congestion control and packet loss during collective bursts. It would not provide the evidence needed for the paired diagnosis and follow-up: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

C:

Define trusted classification boundaries and validate consistent class-to-queue behavior across endpoints and switches. It is not the scoped response supported by the accompanying evidence: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

Question 335

The acceptance board must decide whether an AI-infrastructure response to rocev2 routing prerequisites is fit for controlled operation. The fictional operating facts are: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Signoff cannot rely on a good-looking utilization graph, a change ticket alone or an unmeasured promise. The response must include an appropriate governing engineering principle, a scoped intervention and a separate measurable acceptance criterion. Choose EXACTLY THREE statements that meet those roles.

  1. Changes to loss-sensitive AI traffic need dependency mapping, rollback and independent acceptance measurements.
  2. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.
  3. Correct UCS LAN connectivity policies with staged host-by-host validation and independent uplink paths.
  4. Parallel links carry balanced measured demand and the AI job completes within its performance target.
  5. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.
  6. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

Explanation

Correct Answer

B:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

E:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

F:

A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

Incorrect Answers

A:

Changes to loss-sensitive AI traffic need dependency mapping, rollback and independent acceptance measurements. It is not the rule that supports the proposed change and acceptance test: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

C:

Correct UCS LAN connectivity policies with staged host-by-host validation and independent uplink paths. It does not implement the governing rule for this approval: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

D:

Parallel links carry balanced measured demand and the AI job completes within its performance target. It would not independently demonstrate that the control works: A multi-rack GPU job works inside one subnet but fails large RDMA messages after traversing a routed spine path. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

Question 336

A data-center AI architecture committee considers rocev2 routing prerequisites in its AI Infrastructure Deployment and Data Management review. At the affected site: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. The committee is deliberately not selecting a vendor product or applying a live change yet. It needs the foundational reasoning that makes the next design decision defensible. Which ONE engineering principle directly governs this situation? Choose ONE answer.

  1. ECN should signal incipient congestion so capable endpoints react before queues overflow, with marking thresholds validated against workload traffic.
  2. Intersight-driven compute management depends on authorized targets, compatible profiles and predictable policy inheritance.
  3. UCS vNIC templates must align VLAN or fabric assignment, MTU, adapter policies and unique identity pools.
  4. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

Explanation

Correct Answer

D:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Incorrect Answers

A:

ECN should signal incipient congestion so capable endpoints react before queues overflow, with marking thresholds validated against workload traffic. It does not govern the condition described: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

B:

Intersight-driven compute management depends on authorized targets, compatible profiles and predictable policy inheritance. It does not govern the condition described: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

C:

UCS vNIC templates must align VLAN or fabric assignment, MTU, adapter policies and unique identity pools. It does not govern the condition described: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

Question 337

An infrastructure diagnostician is investigating rocev2 routing prerequisites for the AI Infrastructure Deployment and Data Management team. The reported incident reads: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Competing teams attribute the symptoms to different layers of the AI stack. Before accepting a hypothesis or touching production settings, the investigator must choose a discriminating set of measurements. Which ONE evidence package is MOST relevant? Choose ONE answer.

  1. Examine proposed diff, current device state, execution identity, idempotence report and rollback plan.
  2. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.
  3. Capture baseline queue health, job completion time, path MTU, host/leaf settings, change-window dependencies and rollback triggers.
  4. Inspect class mapping, offered traffic rate, bandwidth allocation and starvation on storage and control classes.

Explanation

Correct Answer

B:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

Incorrect Answers

A:

Examine proposed diff, current device state, execution identity, idempotence report and rollback plan. That check would not establish the decisive facts for this condition: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

C:

Capture baseline queue health, job completion time, path MTU, host/leaf settings, change-window dependencies and rollback triggers. That check would not establish the decisive facts for this condition: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

D:

Inspect class mapping, offered traffic rate, bandwidth allocation and starvation on storage and control classes. That check would not establish the decisive facts for this condition: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Question 338

During a bounded AI-platform change-control review concerning rocev2 routing prerequisites, the site owner reports: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. The proposed work must have a narrow implementation scope, first be validated against the failure evidence and avoid unrelated system-wide tuning. Which ONE intervention should the engineers pilot before any wider rollout? Choose ONE answer.

  1. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.
  2. Scope PFC only to validated loss-sensitive priorities and tune headroom while preventing congestion spreading to unrelated flows.
  3. Apply a version-compatible storage or boot policy and test both normal boot and controlled path loss.
  4. Align supported UCS QoS classes with end-to-end data-center QoS policy and test contention.

Explanation

Correct Answer

A:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.

Incorrect Answers

B:

Scope PFC only to validated loss-sensitive priorities and tune headroom while preventing congestion spreading to unrelated flows. That change would leave the reported problem unresolved: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

C:

Apply a version-compatible storage or boot policy and test both normal boot and controlled path loss. That change would leave the reported problem unresolved: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

D:

Align supported UCS QoS classes with end-to-end data-center QoS policy and test contention. That change would leave the reported problem unresolved: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

Question 339

Two independent technical reviewers examine a AI Infrastructure Deployment and Data Management problem involving rocev2 routing prerequisites. Their shared incident log states: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. One reviewer requires measurements capable of separating the observed problem from neighboring failure modes; the other requires a bounded action that follows those measurements. Choose EXACTLY TWO statements that together supply those two parts.

  1. Verify onboarded site inventory, supported features, access roles, software versions and configuration diffs.
  2. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.
  3. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss.
  4. Choose supported load-distribution behavior and adjust flow placement or links without creating packet reordering hazards.
  5. Inspect class mapping, offered traffic rate, bandwidth allocation and starvation on storage and control classes.

Explanation

Correct Answer

B:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

C:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

Incorrect Answers

A:

Verify onboarded site inventory, supported features, access roles, software versions and configuration diffs. It would not provide the evidence needed for the paired diagnosis and follow-up: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

D:

Choose supported load-distribution behavior and adjust flow placement or links without creating packet reordering hazards. It is not the scoped response supported by the accompanying evidence: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

E:

Inspect class mapping, offered traffic rate, bandwidth allocation and starvation on storage and control classes. It would not provide the evidence needed for the paired diagnosis and follow-up: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Verify source and destination NIC RoCE mode, IP reachability, path MTU, DSCP/priority mapping and observed retransmission or loss. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

Question 340

The acceptance board must decide whether an AI-infrastructure response to rocev2 routing prerequisites is fit for controlled operation. The fictional operating facts are: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Signoff cannot rely on a good-looking utilization graph, a change ticket alone or an unmeasured promise. The response must include an appropriate governing engineering principle, a scoped intervention and a separate measurable acceptance criterion. Choose EXACTLY THREE statements that meet those roles.

  1. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.
  2. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.
  3. Time synchronization across UCS, fabric, GPU hosts and monitoring is needed for reliable correlation and operational security.
  4. Maximum approved training load stays within electrical limits without involuntary power capping or host resets.
  5. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.
  6. Configure supported ETS class weights to protect essential shares without starving management and storage traffic.

Explanation

Correct Answer

A:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

B:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths.

E:

A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

Incorrect Answers

C:

Time synchronization across UCS, fabric, GPU hosts and monitoring is needed for reliable correlation and operational security. It is not the rule that supports the proposed change and acceptance test: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. RoCEv2 uses routable IP/UDP transport but requires consistent fabric MTU, QoS marking and endpoint support across its paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

D:

Maximum approved training load stays within electrical limits without involuntary power capping or host resets. It would not independently demonstrate that the control works: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic.

F:

Configure supported ETS class weights to protect essential shares without starving management and storage traffic. It does not implement the governing rule for this approval: A storage acceleration pod uses mixed RDMA NIC profiles and some workers silently fall back to slower paths. Align endpoint and fabric settings on a documented RoCEv2 path before high-rate collective traffic. Cross-subnet RDMA tests meet loss and latency requirements without fragmentation or class mismatch.

Implementing Cisco Data Center AI Infrastructure (DCAI) (300-640) Practice Tests