mcAfee Secure Website
exam =5
exam =6

Exam Code: JN0-481

Exam Name: Data Center, Specialist (JNCIS-DC)

Certification Provider: Juniper

Juniper JN0-481 Questions & Answers

Reliable & Actual Study Materials for JN0-481 Exam Success

65 Questions & Answers with Testing Engine

"JN0-481: Data Center, Specialist (JNCIS-DC)" Testing Engine covers all the knowledge points of the real Juniper JN0-481 exam.

The latest actual JN0-481 Questions & Answers from Pass4sure. Everything you need to prepare and get best score at JN0-481 exam easily and quickly.

exam =7
Guarantee

Satisfaction Guaranteed

Pass4sure has a remarkable Juniper Candidate Success record. We're confident of our products and provide no hassle product exchange. That's how confident we are!

99.3% Pass Rate
Was: $137.49
Now: $124.99

Juniper JN0-481 Practice Test 44

Question 430

A built-in analytics probe reports a metric but cannot identify why the value changed after a rack migration. Which shortcut would be most likely to undermine this objective?

  1. Raise the anomaly threshold to infinity because the probe currently selects unrelated devices.
  2. Trigger incidents from a single unvalidated stale sample while discarding the source telemetry.
  3. Disable interface telemetry to suppress alerts caused by rising packet errors.
  4. Treat absence of a service configuration anomaly as proof that every application performance metric is normal.

Explanation

Correct Answer

A:

This is the unsafe shortcut to avoid: raise the anomaly threshold to infinity because the probe currently selects unrelated devices. It violates the principle that an IBA probe's context and graph selector are essential to interpreting its measurements and anomaly scope. An appropriate alternative is to narrow the query and probe stages to the correct blueprint nodes and monitored metrics.

Incorrect Answers

B:

“Trigger incidents from a single unvalidated stale sample while discarding the source telemetry” is associated with a distinct issue: A team wants an analytics alert if interface utilization deviates from baseline while still retaining raw evidence. By contrast, the actual problem is: A built-in analytics probe reports a metric but cannot identify why the value changed after a rack migration. An IBA probe's context and graph selector are essential to interpreting its measurements and anomaly scope. The directly relevant unsafe shortcut is “Raise the anomaly threshold to infinity because the probe currently selects unrelated devices”.

C:

The option “Disable interface telemetry to suppress alerts caused by rising packet errors” is useful in a different setting: An operator wants an early-warning signal when fabric interface errors exceed a defined threshold. The case being tested is: A built-in analytics probe reports a metric but cannot identify why the value changed after a rack migration. An IBA probe's context and graph selector are essential to interpreting its measurements and anomaly scope. The directly relevant unsafe shortcut is “Raise the anomaly threshold to infinity because the probe currently selects unrelated devices”.

D:

“Treat absence of a service configuration anomaly as proof that every application performance metric is normal” is relevant to a different condition: An IBA probe reports intermittent latency spikes but application errors are not visible in basic device health summaries. The stated question instead concerns: A built-in analytics probe reports a metric but cannot identify why the value changed after a rack migration. An IBA probe's context and graph selector are essential to interpreting its measurements and anomaly scope. The directly relevant unsafe shortcut is “Raise the anomaly threshold to infinity because the probe currently selects unrelated devices”.

Question 431

A blueprint uses a template intended for a different network OS and generated configuration contains unsupported syntax. Which controlled change best addresses the described requirement?

  1. Resolve the specific design and resource conflicts and regenerate intended configuration.
  2. Choose an officially supported template and platform combination and regenerate the blueprint configuration.
  3. Stage and review each intended blueprint update through an approved migration or change process.
  4. Extend the approved blueprint design to include appropriate border connectivity and associated resources.

Explanation

Correct Answer

B:

This action addresses the stated defect: choose an officially supported template and platform combination and regenerate the blueprint configuration. The underlying constraint is that blueprint templates, device profiles and NOS-specific rendering must align to the supported combination. Verify rather than assume success: verify syntax validation and successful controlled deployment to representative supported devices.

Incorrect Answers

A:

“Resolve the specific design and resource conflicts and regenerate intended configuration” is relevant to a different condition: A new blueprint shows red validation errors before any managed device has received configuration. The stated question instead concerns: A blueprint uses a template intended for a different network OS and generated configuration contains unsupported syntax. Choose an officially supported template and platform combination and regenerate the blueprint configuration. The most relevant correction is “Choose an officially supported template and platform combination and regenerate the blueprint configuration”.

C:

“Stage and review each intended blueprint update through an approved migration or change process” is associated with a distinct issue: A reference template changes its interface model after three production blueprints have already deployed. By contrast, the actual problem is: A blueprint uses a template intended for a different network OS and generated configuration contains unsupported syntax. Choose an officially supported template and platform combination and regenerate the blueprint configuration. The most relevant correction is “Choose an officially supported template and platform combination and regenerate the blueprint configuration”.

D:

The option “Extend the approved blueprint design to include appropriate border connectivity and associated resources” is useful in a different setting: An engineer finishes a fabric template without any modeled north-south peering attachment points. The case being tested is: A blueprint uses a template intended for a different network OS and generated configuration contains unsupported syntax. Choose an officially supported template and platform combination and regenerate the blueprint configuration. The most relevant correction is “Choose an officially supported template and platform combination and regenerate the blueprint configuration”.

Question 432

An automation needs to find only leafs assigned to a particular rack, but a free-text search returns similarly named unrelated devices. Which design or operational principle best explains this behavior?

  1. Blueprint object tags and query filters support targeted discovery and operational reporting.
  2. Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches.
  3. An Apstra service anomaly indicates observed service state deviates from modeled intent and need not indicate a physical interface failure.
  4. Reverting uncommitted edits is distinct from using Time Voyager to select a previously committed blueprint revision.

Explanation

Correct Answer

B:

blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches. Here, an automation needs to find only leafs assigned to a particular rack, but a free-text search returns similarly named unrelated devices. A useful discriminating check is to inspect object types, graph relationship properties and returned identities for the intended rack.

Incorrect Answers

A:

The option “Blueprint object tags and query filters support targeted discovery and operational reporting” is useful in a different setting: A team wants to locate every virtual network owned by a particular application during a migration. The case being tested is: An automation needs to find only leafs assigned to a particular rack, but a free-text search returns similarly named unrelated devices. Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches. The controlling principle is “Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches”.

C:

That choice—“An Apstra service anomaly indicates observed service state deviates from modeled intent and need not indicate a physical interface failure”—addresses this other problem: A blueprint raises a service anomaly for a virtual network even though all interfaces in its rack are administratively up. Here the scenario is: An automation needs to find only leafs assigned to a particular rack, but a free-text search returns similarly named unrelated devices. Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches. The controlling principle is “Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches”.

D:

“Reverting uncommitted edits is distinct from using Time Voyager to select a previously committed blueprint revision” is associated with a distinct issue: A designer notices an accidental rack parameter change before any blueprint commit has occurred. By contrast, the actual problem is: An automation needs to find only leafs assigned to a particular rack, but a free-text search returns similarly named unrelated devices. Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches. The controlling principle is “Blueprint graph and object queries can select devices using modeled relationships rather than fragile substring matches”.

Question 433

A template identifies uplink ports by logical role but a new hardware SKU has different port naming. Which controlled change best addresses the described requirement?

  1. Reallocate the conflicting interface roles to separate supported physical ports.
  2. Choose a compatible interface map or adjust the approved breakout and device profile.
  3. Apply a supported interface map that binds each logical fabric role to the intended physical port.
  4. Stage and review each intended blueprint update through an approved migration or change process.

Explanation

Correct Answer

C:

This action addresses the stated defect: apply a supported interface map that binds each logical fabric role to the intended physical port. The underlying constraint is that logical device interfaces must be mapped to actual supported physical device ports through the correct interface map. Verify rather than assume success: confirm rendered port assignments match the physical cable map and supported switch interfaces.

Incorrect Answers

A:

“Reallocate the conflicting interface roles to separate supported physical ports” is associated with a distinct issue: A fabric build fails because one interface was assigned to both an uplink and a server-facing logical port. By contrast, the actual problem is: A template identifies uplink ports by logical role but a new hardware SKU has different port naming. The governing requirement is to apply a supported interface map that binds each logical fabric role to the intended physical port. The most relevant correction is “Apply a supported interface map that binds each logical fabric role to the intended physical port”.

B:

The option “Choose a compatible interface map or adjust the approved breakout and device profile” is useful in a different setting: An Apstra design refuses a proposed uplink because its interface map does not match hardware port mode. The case being tested is: A template identifies uplink ports by logical role but a new hardware SKU has different port naming. The governing requirement is to apply a supported interface map that binds each logical fabric role to the intended physical port. The most relevant correction is “Apply a supported interface map that binds each logical fabric role to the intended physical port”.

D:

That choice—“Stage and review each intended blueprint update through an approved migration or change process”—addresses this other problem: A reference template changes its interface model after three production blueprints have already deployed. Here the scenario is: A template identifies uplink ports by logical role but a new hardware SKU has different port naming. The governing requirement is to apply a supported interface map that binds each logical fabric role to the intended physical port. The most relevant correction is “Apply a supported interface map that binds each logical fabric role to the intended physical port”.

Question 434

An operator sees devices in mixed managed states and cannot tell which switches received the new active configuration. Which design or operational principle best explains this behavior?

  1. Cabling validation compares observed neighbor links against intended device/interface relationships.
  2. Cable-map discrepancies should be reconciled before traffic flows through an unvalidated physical fabric.
  3. Controlled build and deploy phases separate design validation, reviewed staged changes and active device updates.
  4. Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly.

Explanation

Correct Answer

D:

per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly. Here, an operator sees devices in mixed managed states and cannot tell which switches received the new active configuration. A useful discriminating check is to check each device's state, failed action details and intended-versus-running configuration revision.

Incorrect Answers

A:

That choice—“Cabling validation compares observed neighbor links against intended device/interface relationships”—addresses this other problem: A live rack cable is connected to the wrong spine port compared with the blueprint's intended cable map. Here the scenario is: An operator sees devices in mixed managed states and cannot tell which switches received the new active configuration. Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly. The controlling principle is “Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly”.

B:

“Cable-map discrepancies should be reconciled before traffic flows through an unvalidated physical fabric” is associated with a distinct issue: A staging team has different cable-map snapshots from design and observed discovery data. By contrast, the actual problem is: An operator sees devices in mixed managed states and cannot tell which switches received the new active configuration. Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly. The controlling principle is “Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly”.

C:

The option “Controlled build and deploy phases separate design validation, reviewed staged changes and active device updates” is useful in a different setting: An Apstra operator wants to distinguish blueprint validation from the actual deployment to managed switches. The case being tested is: An operator sees devices in mixed managed states and cannot tell which switches received the new active configuration. Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly. The controlling principle is “Per-device deployment and configuration status must be inspected rather than assuming a blueprint commit succeeded uniformly”.

Question 435

A newly created virtual network must be attached to a tenant VRF, but application packets route through the default routing instance. Which evidence most directly verifies the intended network outcome?

  1. Confirm the application subnet uses the expected tenant VRF and other tenants remain isolated.
  2. Confirm each tenant uses its own routing table and no unapproved inter-tenant connectivity occurs.
  3. Confirm both VN endpoints are isolated at Layer 2 but routed according to tenant policy within the VRF.
  4. Verify permitted shared-service traffic works and all other inter-tenant paths remain blocked.

Explanation

Correct Answer

A:

This directly tests the desired result: confirm the application subnet uses the expected tenant VRF and other tenants remain isolated. The decisive investigation is to inspect the VN's routing-zone binding, SVI and VRF route lookup on the affected leaf. The governing principle is that a virtual network's routed gateway and routing-zone association determine its Layer 3 tenant boundary.

Incorrect Answers

B:

That choice—“Confirm each tenant uses its own routing table and no unapproved inter-tenant connectivity occurs”—addresses this other problem: Two tenants require identical subnet ranges in a shared physical fabric while remaining isolated. Here the scenario is: A newly created virtual network must be attached to a tenant VRF, but application packets route through the default routing instance. A virtual network's routed gateway and routing-zone association determine its Layer 3 tenant boundary. The direct acceptance evidence is “Confirm the application subnet uses the expected tenant VRF and other tenants remain isolated”.

C:

“Confirm both VN endpoints are isolated at Layer 2 but routed according to tenant policy within the VRF” is associated with a distinct issue: A user asks whether a single tenant routing zone can include multiple Layer 2 VNs attached to different racks. By contrast, the actual problem is: A newly created virtual network must be attached to a tenant VRF, but application packets route through the default routing instance. A virtual network's routed gateway and routing-zone association determine its Layer 3 tenant boundary. The direct acceptance evidence is “Confirm the application subnet uses the expected tenant VRF and other tenants remain isolated”.

D:

The option “Verify permitted shared-service traffic works and all other inter-tenant paths remain blocked” is useful in a different setting: Two tenants require a controlled shared service, and a proposal suggests importing every route into each tenant VRF. The case being tested is: A newly created virtual network must be attached to a tenant VRF, but application packets route through the default routing instance. A virtual network's routed gateway and routing-zone association determine its Layer 3 tenant boundary. The direct acceptance evidence is “Confirm the application subnet uses the expected tenant VRF and other tenants remain isolated”.

Question 436

A new leaf model arrives with an unsupported operating-system image while the Apstra server still shows available inventory capacity. Which design or operational principle best explains this behavior?

  1. Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation.
  2. The Apstra server relies on managed device agent connectivity and state collection to maintain an accurate operational model.
  3. The controller's intended blueprint state and deployed device configurations must be reconciled after an out-of-band restore.
  4. Centralized intent and change management should distinguish staged edits, reviewed changes and active deployment ownership.

Explanation

Correct Answer

A:

supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation. Here, a new leaf model arrives with an unsupported operating-system image while the Apstra server still shows available inventory capacity. A useful discriminating check is to check supported platform and NOS versions alongside device profile and agent compatibility before assigning the switch.

Incorrect Answers

B:

“The Apstra server relies on managed device agent connectivity and state collection to maintain an accurate operational model” is relevant to a different condition: A control-plane service is healthy yet one switch disappears from the Apstra device inventory after a management-network change. The stated question instead concerns: A new leaf model arrives with an unsupported operating-system image while the Apstra server still shows available inventory capacity. Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation. The controlling principle is “Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation”.

C:

That choice—“The controller's intended blueprint state and deployed device configurations must be reconciled after an out-of-band restore”—addresses this other problem: An operations team recovers a backup of the controller taken before the most recent fabric changes. Here the scenario is: A new leaf model arrives with an unsupported operating-system image while the Apstra server still shows available inventory capacity. Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation. The controlling principle is “Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation”.

D:

“Centralized intent and change management should distinguish staged edits, reviewed changes and active deployment ownership” is associated with a distinct issue: A site adds a second blueprint operator but needs to ensure simultaneous edits do not bypass change ownership. By contrast, the actual problem is: A new leaf model arrives with an unsupported operating-system image while the Apstra server still shows available inventory capacity. Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation. The controlling principle is “Supported device integration requires compatible platform profiles, agents and operating-system capabilities for intended automation”.

Question 437

An audit asks why some device commands come from the baseline while others appear from intended virtual-network configuration. Which shortcut would be most likely to undermine this objective?

  1. Treat unmanaged server ports as invisible and configure arbitrary switch interfaces by hand.
  2. Treat every difference between a pristine config and final rendered config as an unauthorized attack.
  3. Create the new rack outside the blueprint so its configurations cannot be tracked centrally.
  4. Paste untracked manual commands into every switch and rely on Apstra never to overwrite them.

Explanation

Correct Answer

B:

This is the unsafe shortcut to avoid: treat every difference between a pristine config and final rendered config as an unauthorized attack. It violates the principle that configuration views and their origins should be compared before diagnosing managed-device drift. An appropriate alternative is to investigate unexpected changes using the appropriate generated-config view and approved override workflow.

Incorrect Answers

A:

“Treat unmanaged server ports as invisible and configure arbitrary switch interfaces by hand” is associated with a distinct issue: A new dual-homed application server is physically connected but its logical attachment is absent from the active blueprint. By contrast, the actual problem is: An audit asks why some device commands come from the baseline while others appear from intended virtual-network configuration. Configuration views and their origins should be compared before diagnosing managed-device drift. The directly relevant unsafe shortcut is “Treat every difference between a pristine config and final rendered config as an unauthorized attack”.

C:

“Create the new rack outside the blueprint so its configurations cannot be tracked centrally” is relevant to a different condition: A data center needs to add another rack to a production fabric without replacing the existing blueprint. The stated question instead concerns: An audit asks why some device commands come from the baseline while others appear from intended virtual-network configuration. Configuration views and their origins should be compared before diagnosing managed-device drift. The directly relevant unsafe shortcut is “Treat every difference between a pristine config and final rendered config as an unauthorized attack”.

D:

That choice—“Paste untracked manual commands into every switch and rely on Apstra never to overwrite them”—addresses this other problem: A proprietary monitoring agent needs device-level CLI configuration not generated by standard Apstra intents. Here the scenario is: An audit asks why some device commands come from the baseline while others appear from intended virtual-network configuration. Configuration views and their origins should be compared before diagnosing managed-device drift. The directly relevant unsafe shortcut is “Treat every difference between a pristine config and final rendered config as an unauthorized attack”.

Question 438

A team investigates whether a blueprint revision was created by automation or by an administrator. Which initial check would most directly distinguish the likely cause?

  1. Review the event timeline, blueprint commit notes, deployment status and actor records.
  2. Correlate blueprint revision history, actor identity and event logs around the failure timestamp.
  3. Inspect supported syslog configuration, transport reachability and receiving collector parsing.
  4. Check collector or agent timestamps, sample rates and ingestion connectivity before trusting status.

Explanation

Correct Answer

B:

This checks the relevant evidence first: correlate blueprint revision history, actor identity and event logs around the failure timestamp. The underlying concept is that Apstra event and audit records provide change accountability distinct from raw packet telemetry. After confirming the finding, a controlled remediation is to preserve audit records and review the specific committed change before a controlled correction.

Incorrect Answers

A:

That choice—“Review the event timeline, blueprint commit notes, deployment status and actor records”—addresses this other problem: A team needs to identify whether a failed deployment arose from system automation or an approved human change. Here the scenario is: A team investigates whether a blueprint revision was created by automation or by an administrator. Correlate blueprint revision history, actor identity and event logs around the failure timestamp. The most discriminating first check is “Correlate blueprint revision history, actor identity and event logs around the failure timestamp”.

C:

The option “Inspect supported syslog configuration, transport reachability and receiving collector parsing” is useful in a different setting: A data center runs a centralized SOC and requests appliance event forwarding without sharing administrator credentials. The case being tested is: A team investigates whether a blueprint revision was created by automation or by an administrator. Correlate blueprint revision history, actor identity and event logs around the failure timestamp. The most discriminating first check is “Correlate blueprint revision history, actor identity and event logs around the failure timestamp”.

D:

“Check collector or agent timestamps, sample rates and ingestion connectivity before trusting status” is relevant to a different condition: A monitoring dashboard shows a device as green but telemetry timestamps stopped updating 20 minutes earlier. The stated question instead concerns: A team investigates whether a blueprint revision was created by automation or by an administrator. Correlate blueprint revision history, actor identity and event logs around the failure timestamp. The most discriminating first check is “Correlate blueprint revision history, actor identity and event logs around the failure timestamp”.

Question 439

An operator marks IBA alarms resolved by weakening thresholds while the original interface drops continue. Which initial check would most directly distinguish the likely cause?

  1. Inspect affected probe stages, timestamps and source telemetry along the application path.
  2. Group probe anomalies by affected node and stage, then inspect linked telemetry and topology edges.
  3. Compare live raw telemetry, probe thresholds and application SLOs before and after the change.
  4. Inspect probe data sources, monitored stages, thresholds and the exact interfaces being selected.

Explanation

Correct Answer

C:

This checks the relevant evidence first: compare live raw telemetry, probe thresholds and application SLOs before and after the change. The underlying concept is that a cleared threshold alarm alone is not sufficient proof of service recovery; measured data and intent must support the conclusion. After confirming the finding, a controlled remediation is to restore justified monitoring sensitivity and remediate the measured issue instead of hiding it.

Incorrect Answers

A:

“Inspect affected probe stages, timestamps and source telemetry along the application path” is relevant to a different condition: An IBA probe reports intermittent latency spikes but application errors are not visible in basic device health summaries. The stated question instead concerns: An operator marks IBA alarms resolved by weakening thresholds while the original interface drops continue. The relevant inspection is: Compare live raw telemetry, probe thresholds and application SLOs before and after the change. The most discriminating first check is “Compare live raw telemetry, probe thresholds and application SLOs before and after the change”.

B:

That choice—“Group probe anomalies by affected node and stage, then inspect linked telemetry and topology edges”—addresses this other problem: An operations team receives repeated probe-stage alerts after an optic begins incrementing errors. Here the scenario is: An operator marks IBA alarms resolved by weakening thresholds while the original interface drops continue. The relevant inspection is: Compare live raw telemetry, probe thresholds and application SLOs before and after the change. The most discriminating first check is “Compare live raw telemetry, probe thresholds and application SLOs before and after the change”.

D:

The option “Inspect probe data sources, monitored stages, thresholds and the exact interfaces being selected” is useful in a different setting: An operator wants an early-warning signal when fabric interface errors exceed a defined threshold. The case being tested is: An operator marks IBA alarms resolved by weakening thresholds while the original interface drops continue. The relevant inspection is: Compare live raw telemetry, probe thresholds and application SLOs before and after the change. The most discriminating first check is “Compare live raw telemetry, probe thresholds and application SLOs before and after the change”.

Question 440

An operator finds two equal-cost fabric paths but a single large traffic flow consistently traverses only one link. Which shortcut would be most likely to undermine this objective?

  1. Treat the presence of two physical links as proof that routing automatically reconverges instantly.
  2. Add unrestricted lateral links between leaf devices without updating the design model.
  3. Assume ECMP must divide every single packet evenly across all active links.
  4. Create new tenant VNIs to mask an unresolved underlay routing failure.

Explanation

Correct Answer

C:

This is the unsafe shortcut to avoid: assume ECMP must divide every single packet evenly across all active links. It violates the principle that equal-cost multipath commonly hashes individual flows to a path while distributing many flows across available next hops. An appropriate alternative is to validate resilient ECMP and, if required, test multiple flows or adjust workload distribution within approved designs.

Incorrect Answers

A:

The option “Treat the presence of two physical links as proof that routing automatically reconverges instantly” is useful in a different setting: A fabric uses two equal-cost underlay next hops yet packets continue toward a failed path following a port outage. The case being tested is: An operator finds two equal-cost fabric paths but a single large traffic flow consistently traverses only one link. Equal-cost multipath commonly hashes individual flows to a path while distributing many flows across available next hops. The directly relevant unsafe shortcut is “Assume ECMP must divide every single packet evenly across all active links”.

B:

“Add unrestricted lateral links between leaf devices without updating the design model” is relevant to a different condition: A migration plan connects every rack leaf laterally and treats spine devices as only optional edge routers. The stated question instead concerns: An operator finds two equal-cost fabric paths but a single large traffic flow consistently traverses only one link. Equal-cost multipath commonly hashes individual flows to a path while distributing many flows across available next hops. The directly relevant unsafe shortcut is “Assume ECMP must divide every single packet evenly across all active links”.

D:

“Create new tenant VNIs to mask an unresolved underlay routing failure” is associated with a distinct issue: A data center team adds a leaf switch but the new node cannot reach any spine loopback addresses although local links are up. By contrast, the actual problem is: An operator finds two equal-cost fabric paths but a single large traffic flow consistently traverses only one link. Equal-cost multipath commonly hashes individual flows to a path while distributing many flows across available next hops. The directly relevant unsafe shortcut is “Assume ECMP must divide every single packet evenly across all active links”.

Data Center, Specialist (JNCIS-DC) (JN0-481) Practice Tests