AP Networking Network Design and Redundancy Practice Test

1. Any device with one hostname
2. A component whose failure can interrupt the required service without an effective alternate
3. Every device that uses TCP
4. Any cable shorter than one meter

Correct answer: B. A component whose failure can interrupt the required service without an effective alternate

A single point concerns dependency and impact. Count actual independent service paths, not just the number of equipment boxes.

1. All endpoints automatically form a mesh
2. Every endpoint becomes a router
3. Only DNS names change
4. Connected endpoints lose the communication dependent on that switch

Correct answer: D. Connected endpoints lose the communication dependent on that switch

The center is a shared dependency. The result depends on which services and paths use it, so scope the failure effect explicitly.

1. Replace all monitoring
2. Provide an alternate path or capability when a required component fails
3. Remove the need for policy
4. Guarantee zero disruption for every application

Correct answer: B. Provide an alternate path or capability when a required component fails

Redundancy offers alternatives under a defined failure model. Detection, convergence, capacity, and state continuity still determine useful recovery.

1. DNS chooses the switch's forwarding port
2. Uncontrolled Layer 2 loops can replicate traffic and destabilize learning
3. TCP prevents all switching loops
4. Every extra link automatically increases TTL

Correct answer: B. Uncontrolled Layer 2 loops can replicate traffic and destabilize learning

Ethernet forwarding needs a loop-free effective topology or supported aggregation design. Adding links without control can reduce reliability.

1. A public address per port
2. A DNS cache for every switch
3. Application user authentication
4. A loop-free active Layer 2 topology from redundant links

Correct answer: D. A loop-free active Layer 2 topology from redundant links

STP controls which paths forward within its scope. It does not provide every aspect of routed, power, or application redundancy.

1. A set of resources affected together by a particular failure
2. Only a DNS suffix
3. A password-expiry interval
4. Only one public IP address

Correct answer: A. A set of resources affected together by a particular failure

Shared power, devices, ducts, providers, or control systems can define correlated failures. Identify the specific event and dependent resources.

1. Their different IP addresses
2. Their different rack labels
3. That power source and its failure path
4. Their separate hostnames

Correct answer: C. That power source and its failure path

Duplicated devices can fail together when they share power. Independent-looking equipment does not prove independent failure domains.

1. Only the largest VLAN number
2. Only one topology drawing with no requirements
3. Required services, users, constraints, and acceptable failure behavior
4. A random list of the most expensive devices

Correct answer: C. Required services, users, constraints, and acceptable failure behavior

A design is evaluated against requirements. Capacity, security, reliability, and maintainability choices follow the intended tasks and constraints.

1. Only TLS certificate names
2. Only DNS record types
3. Only user permissions
4. Actual devices, ports, cabling, and placement

Correct answer: D. Actual devices, ports, cabling, and placement

Physical documentation describes connections and locations. A logical diagram adds addressing, segmentation, and forwarding relationships that may differ from physical appearance.

1. Only cable lengths
2. Only monitor placement
3. Only rack paint color
4. Addressing, segments, routing, and traffic relationships

Correct answer: D. Addressing, segments, routing, and traffic relationships

Logical relationships show how communication is organized. Two physical links can belong to different logical segments or share a common service dependency.

1. To change its IP version
2. To eliminate all link errors
3. To identify its actual path and support maintenance
4. To make it carry twice the bandwidth

Correct answer: C. To identify its actual path and support maintenance

Endpoint and port labels prevent guessing during repair. Keep the record aligned with the installed path.

1. That every task must stay identical automatically
2. That ping alone proves every requirement
3. Which required tasks remain supported during failover
4. That backup capacity is irrelevant

Correct answer: C. Which required tasks remain supported during failover

Degraded operation may need prioritization or reduced workload. Define and test the intended service level under the backup path.

1. The expiration of a DNS alias
2. The naming of a switch port
3. The encryption of every packet
4. The process of reaching a usable forwarding state after a change

Correct answer: D. The process of reaching a usable forwarding state after a change

Recovery can take time for detection and forwarding updates. Link presence alone does not imply the alternate path is already usable.

1. Every redundant design is self-proving
2. Tests replace all documentation
3. Every failover is harmless regardless of scope
4. Configuration intent may differ from actual recovery behavior

Correct answer: D. Configuration intent may differ from actual recovery behavior

A controlled test reveals detection, dependencies, capacity, and service outcomes. Plan scope, observation, and rollback around the required service.

1. The same IP address on every endpoint
2. Any two cables plugged into any switches
3. Compatible supported configuration at both ends and correct physical membership
4. A shared DNS record only

Correct answer: C. Compatible supported configuration at both ends and correct physical membership

Aggregation depends on the platform and topology design. Arbitrary parallel cables are not automatically one safe logical link.

1. Yes; TCP doubles its maximum port
2. No; a single flow may be assigned to one member
3. Yes; every packet always alternates perfectly
4. Only if DNS returns two answers

Correct answer: B. No; a single flow may be assigned to one member

Many aggregation designs distribute flows by a hash rather than stripe every packet. Aggregate capacity and single-flow throughput are different requirements.

1. Every host needs a /8
2. DHCP creates unlimited addresses
3. More unused addresses automatically increase link speed
4. Future endpoints and segments need a planned allocation path

Correct answer: D. Future endpoints and segments need a planned allocation path

Capacity planning should match realistic growth and segmentation needs. Document allocation assumptions instead of making every subnet arbitrarily large.

1. A promise that no failure will occur
2. A list of unrelated URLs
3. A defined way to restore an acceptable prior state if a change fails
4. A new default password

Correct answer: C. A defined way to restore an acceptable prior state if a change fails

A rollback includes prerequisites, saved configuration, trigger, steps, and verification. It supports controlled maintenance without replacing testing.

1. The complete service path can still fail at the unprotected router
2. The switch's battery protects every remote device
3. Every Internet task is protected by the switch UPS
4. The router receives power through DNS

Correct answer: A. The complete service path can still fail at the unprotected router

Power continuity must cover the required dependency chain. Protecting one component does not establish end-to-end resilience.

1. A VLAN name alone encrypts management
2. Management devices need no routes
3. Guests should always configure switches
4. Administration has a different authorization requirement and impact

Correct answer: D. Administration has a different authorization requirement and impact

Design management paths and identities deliberately. Verify that authorized maintenance works and unauthorized sources remain excluded.

1. Only a vendor quote
2. A test showing whether stated requirements are met
3. Only a topology picture opening
4. Only a device powering on

Correct answer: B. A test showing whether stated requirements are met

Acceptance should cover actual tasks, boundaries, capacity, and defined failure cases. Equipment presence is only one prerequisite.

1. Two local paths can still depend on the same DNS, identity, provider, or application service
2. Only physical links can fail
3. Every cloud service is independent by definition
4. Duplicated cables remove every upstream dependency

Correct answer: A. Two local paths can still depend on the same DNS, identity, provider, or application service

Reliability depends on the complete service chain. Identify shared dependencies beyond local routing devices.

1. Component duplication without independent power delivery
2. A routing metric error
3. Two independent failure domains
4. Guaranteed power resilience

Correct answer: A. Component duplication without independent power delivery

Dual supplies can tolerate a supply failure, but a shared circuit can defeat both. Specify which power failures the design must survive.

1. Only a photo of its front panel
2. Only done
3. Only the new device's brand
4. Intent, scope, dependencies, configuration, timing, verification, and rollback

Correct answer: D. Intent, scope, dependencies, configuration, timing, verification, and rollback

A usable record lets another technician understand the decision and recover later. Tie it to the affected services and measured outcomes.

1. A powered device may still fail to deliver the required task
2. Applications never depend on networks
3. Only uptime can be measured
4. They always measure exactly the same thing

Correct answer: A. A powered device may still fail to deliver the required task

Service availability includes routes, policy, identity, endpoints, and dependencies. A live device or interface is not an end-to-end result.

1. Only a DNS typo on one laptop
2. Only one user's password expiry
3. Damage to that shared physical route
4. Only one server's browser cache

Correct answer: C. Damage to that shared physical route

Provider diversity does not prove route diversity. A shared duct or building entrance is a correlated physical failure domain.

1. Failure of the upstream chassis
2. Failure of one uplink member
3. One uplink cable disconnect
4. One member port failure under a working alternate design

Correct answer: A. Failure of the upstream chassis

Parallel links can reduce link-level exposure while retaining a shared device dependency. State which failure classes the topology covers.

1. No; it may be the planned alternate that prevents a loop
2. Yes; every redundant link must forward independently
3. Only if the link carries DNS
4. Yes; spanning tree always disables the entire switch

Correct answer: A. No; it may be the planned alternate that prevents a loop

A nonforwarding role can be intentional. Verify topology, instance scope, and recovery behavior before treating the alternate as wasted or enabling a loop.

1. Backup addressing, inbound routing/publication, and provider support
2. Only one successful outbound ping
3. Only the LAN switch name
4. Only the clients' private DHCP reservations

Correct answer: A. Backup addressing, inbound routing/publication, and provider support

Outbound and inbound requirements can differ. Failover may change public identity or reachability, so publication needs an explicit supported design.

1. Only cable color
2. Only DNS page titles
3. Only the maximum private address
4. Neighbor update and traffic recovery under the actual redundancy implementation

Correct answer: D. Neighbor update and traffic recovery under the actual redundancy implementation

A gateway transition must become usable to clients. Measure adjacency updates and service continuity, not just which device claims an active role.

1. Distinct addressing under the plan
2. Separate broadcast domains
3. Different segment labels
4. The required intersegment access boundary

Correct answer: D. The required intersegment access boundary

Segmentation and authorization are related but separate. Routing can connect every segment unless policy enforces the intended restrictions.

1. Whether DNS can restore missing physical capacity
2. Whether every flow must be exactly 700 Mb/s
3. Whether two configured member names still imply 2 Gb/s
4. Whether surviving capacity and traffic priorities meet requirements

Correct answer: D. Whether surviving capacity and traffic priorities meet requirements

The surviving member cannot provide the previous aggregate capacity. Recovery must be evaluated under degraded demand rather than by link-up status alone.

1. Growth affects only link speed
2. A /26 has 62 ordinary usable host addresses, so the stated demand exceeds it
3. A /26 has 126 usable addresses
4. A /26 always supplies 256 usable addresses

Correct answer: B. A /26 has 62 ordinary usable host addresses, so the stated demand exceeds it

A /26 contains 64 addresses, conventionally excluding network and broadcast leaves 62. Plan addressing and segment needs rather than overrun the allocation.

1. Only the IPv4 prefix
2. Only the uplink's VLAN name
3. The power budget or supported device allocation/design
4. Only the DNS resolver

Correct answer: C. The power budget or supported device allocation/design

Aggregate PoE capacity is separate from port count and data rate. Account for actual device requirements and approved reserve.

1. More ports guarantee expandability
2. Physical facility capacity was omitted
3. DNS naming was too detailed
4. All network ports carry power automatically

Correct answer: B. Physical facility capacity was omitted

Growth includes facilities, power, cooling, cabling, and maintainability. Logical port capacity alone is not a complete expansion plan.

1. Only the active router's rule text
2. Configuration/policy consistency and task success on the backup path
3. Only the backup link LED
4. Only the number of installed fans

Correct answer: B. Configuration/policy consistency and task success on the backup path

An alternate path includes enforcement as well as routing. Keep required policy aligned and test the actual service during failover.

1. Monitoring always survives because it has software
2. A private address ensures independent reporting
3. The outage can remove both service and reporting path
4. The failed switch must still send every alert

Correct answer: C. The outage can remove both service and reporting path

Common infrastructure can hide the event from the collector. Consider an independent check or detection path appropriate to the required failure model.

1. A longer subnet mask only
2. Proof that any cable test covers all failures
3. A second DNS alias only
4. A test or justified analysis of the larger shared failure domain

Correct answer: D. A test or justified analysis of the larger shared failure domain

The tested event is narrower than the claimed resilience. Match validation to the specific components and correlated failures in the requirement.

1. Only rename the client
2. Reconcile documentation with effective configuration and verify the required segment path
3. Only redraw the VLAN label
4. Only increase public bandwidth

Correct answer: B. Reconcile documentation with effective configuration and verify the required segment path

A diagram is intention or record, not live proof. Trace actual attachments and test the intended service before accepting the topology.

1. Effective route preference/metrics and policy for the actual destination
2. Only the cable labels
3. Only DHCP scope names
4. Only the application's icon

Correct answer: A. Effective route preference/metrics and policy for the actual destination

The installed decision controls forwarding. Compare actual route selection with the intended normal and alternate path behavior.

1. Yes; redundant devices guarantee instant recovery
2. Only if the service uses UDP
3. Yes; each step is below 30 seconds
4. No; measured recovery is at least 35 seconds before any additional application delay

Correct answer: D. No; measured recovery is at least 35 seconds before any additional application delay

Serial stages add under the stated model. Include end-to-end task recovery rather than judging each component duration independently.

1. Only a new VLAN ID
2. Application retry/transaction semantics and safe recovery
3. Only interface speed
4. Only cable redundancy

Correct answer: B. Application retry/transaction semantics and safe recovery

The network can recover while application state remains problematic. Define continuity and retry expectations with the service owner.

1. Every user-facing task necessarily
2. The jump host for the defined administration task
3. Only the Ethernet frame size
4. Only DNS TTL

Correct answer: B. The jump host for the defined administration task

Dependency scope matters. The jump host can be a management single point without necessarily being in ordinary data forwarding.

1. 25
2. 10
3. 5
4. 20

Correct answer: B. 10

Each node connects to four others, giving 5 times 4 divided by 2 equals 10 links. Count each physical pair once.

1. 6
2. 1
3. 10
4. 5

Correct answer: D. 5

The new node connects to each of the five existing nodes. The total rises from 10 to 15 links, illustrating growth in cabling/ports.

1. The subnet mask stores the backup
2. DNS recreates all configuration
3. Backups always survive hardware failure internally
4. The configuration record may be unavailable when needed

Correct answer: D. The configuration record may be unavailable when needed

Keep recoverable configuration under an approved independent storage and access process. A local-only copy shares the device failure domain.

1. Every provider core router necessarily
2. The single access circuit and its path
3. Only the application certificate
4. Only one user's browser cache

Correct answer: B. The single access circuit and its path

Provider-side resilience does not eliminate the site's sole attachment. Trace the end-to-end path and contracted failure coverage.

1. Only adding another router logo to the diagram
2. Authorized management and recovery access on the alternate device
3. Only turning off all access controls
4. Only making the VLAN number larger

Correct answer: B. Authorized management and recovery access on the alternate device

Maintainability needs a usable approved control path. Verify that maintenance can occur without broadening administrative access.

1. Only the largest advertised port speed
2. Only the number of new boxes purchased
3. Only the newest model year
4. The service risk, observed dependency, requirements, and actual coverage of each option

Correct answer: D. The service risk, observed dependency, requirements, and actual coverage of each option

Reliability spending should address demonstrated risks and stated goals. Compare coverage, constraints, and remaining common dependencies.

1. No independent host-level resilience
2. Complete resolver redundancy
3. A new application authentication factor
4. A guaranteed alternate physical path

Correct answer: A. No independent host-level resilience

Two addresses can point to one underlying dependency. Verify actual hosting and upstream dependencies before counting independent services.

1. Survival of the shared handoff or power failure
2. Claim handoff-failure tolerance solely from passing two cable tests
3. Treat the two provider labels as proof of independent physical termination
4. Count device and circuit duplication as proof of independent power delivery

Correct answer: A. Survival of the shared handoff or power failure

The tests demonstrate only the failure conditions exercised. A common handoff remains a single dependency for both circuit paths under that event.

1. A forwarding loop with traffic replication and unstable learning
2. Expect automatic single-flow capacity doubling without a supported topology
3. Assume every parallel Ethernet cable can safely forward independently
4. Treat TCP transport recovery as a substitute for Layer 2 loop prevention

Correct answer: A. A forwarding loop with traffic replication and unstable learning

The alternate was controlled to preserve a loop-free topology. Safe additional capacity requires a supported topology and protocol design rather than overriding the protection blindly.

1. End-to-end session continuity across the actual enforcement path
2. Assume a usable route transfers firewall state to every alternate device
3. Test only new connections and promise established-session preservation
4. Declare complete session continuity solely from gateway IP recovery

Correct answer: A. End-to-end session continuity across the actual enforcement path

Route/gateway recovery and state continuity are separate requirements. Test session behavior through the real alternate devices and document acceptable reconnection semantics.

1. Use the primary capacity in the backup demand comparison
2. Preserve required workloads with approved scheduling/rate treatment of optional transfers and verify combined behavior
3. Validate only idle echo replies before approving busy degraded operation
4. Promise the unchanged 85 Mb/s workload fits because the backup link is up

Correct answer: B. Preserve required workloads with approved scheduling/rate treatment of optional transfers and verify combined behavior

The aggregate demand exceeds the backup capacity. An explicit degraded mode can prioritize required tasks, but must be tested for delay, overhead, and actual performance.

1. Count duplicate devices instead of tracing the user's dependency chain
2. Service-level dependency analysis beyond duplicated network hardware
3. Exclude the shared identity provider because every router has a duplicate
4. Assume working IP forwarding guarantees application authorization during provider loss

Correct answer: B. Service-level dependency analysis beyond duplicated network hardware

The user task depends on identity as well as forwarding. Include the provider and its recovery/degraded behavior in the failure model.

1. Verify only the primary API transaction and extrapolate to the alternate path
2. The real approved API transaction over the alternate egress
3. Assume a changed public source cannot affect remote authorization
4. Use one successful gateway ping as the complete API failover test

Correct answer: B. The real approved API transaction over the alternate egress

Alternate reachability can still violate a remote access condition. Coordinate supported source identities and verify the user-facing operation under failover.

1. Count separate fiber pairs as separate equipment-room failure domains
2. Declare all facility failures covered because fiber routes are geographically separate
3. Ignore the shared termination room's power and environmental dependencies
4. Failure of the shared termination location or facilities

Correct answer: D. Failure of the shared termination location or facilities

Route diversity covers some physical events but not all shared endpoints. Define the room-level power, fire, access, and equipment dependencies relevant to the design.

1. Treat all 128 /25 addresses as ordinary host slots and ignore projected demand
2. Approve the /25 because only the current 100 endpoints are connected
3. The 143-host demand exceeds 126 ordinary usable addresses; plan an appropriate segment/address expansion
4. Address only link bandwidth and leave the exceeded host allocation unchanged

Correct answer: C. The 143-host demand exceeds 126 ordinary usable addresses; plan an appropriate segment/address expansion

Include projected endpoints and infrastructure in capacity. A /25 contains 128 addresses with 126 conventional host slots, so current fit does not establish future fit.

1. Assume flow hashing always alternates consecutive packets equally
2. One flow may remain limited to one member under the supported distribution model
3. Verify aggregate many-flow throughput and treat it as single-flow acceptance
4. Guarantee 2 Gb/s to one flow solely from the two member ratings

Correct answer: B. One flow may remain limited to one member under the supported distribution model

Aggregate throughput under many flows does not guarantee one-flow capacity. Match the actual workload and platform behavior to the requirement.

1. Model-aware mapping and verification of effective attachments
2. Reuse old interface numbers without checking the replacement platform
3. Verify only management login and omit endpoint segment attachments
4. Accept restored configuration text as proof that physical port roles match

Correct answer: A. Model-aware mapping and verification of effective attachments

Configuration syntax restoration is not the same as physical/logical equivalence. Map ports and supported features, then verify required paths and restrictions.

1. It fails the goal with 32 seconds to task recovery
2. Omit application reconnect time from a service-recovery requirement
3. Compare each serial stage separately with the total 25-second task target
4. Accept the design because forwarding convergence alone is four seconds

Correct answer: A. It fails the goal with 32 seconds to task recovery

Acceptance concerns the task's full interruption. Serial stages add to 32 seconds, so shorten a relevant stage or revise the approved requirement with evidence.

1. Treat a saved file on the unreachable device as executable rollback
2. Define rollback steps but omit required access and credentials
3. An approved recovery/control path available under that failure condition
4. Rely on the same failed management path without an alternate recovery method

Correct answer: C. An approved recovery/control path available under that failure condition

A rollback must be executable when needed. Verify independent or supported out-of-band access and prerequisites rather than equate file existence with recoverability.

1. Treat different provider names as proof of separate last-mile carriers
2. Ignore shared ducts because each circuit has a distinct logical identifier
3. Actual shared physical and operational dependencies against the required failure classes
4. Use two invoices as conclusive physical-path independence

Correct answer: C. Actual shared physical and operational dependencies against the required failure classes

Commercial distinction is not physical independence. Obtain the relevant path and handoff evidence and state what diversity the contract/design actually provides.

1. Ignore identity dependencies because the network interfaces remained up
2. Representative load, identity dependencies, policy, and task performance in degraded mode
3. Approve 200-user degraded performance from one idle ping
4. Use the primary link's capacity as the backup workload limit

Correct answer: B. Representative load, identity dependencies, policy, and task performance in degraded mode

Idle reachability is a narrow prerequisite. Acceptance needs the workload and service conditions named in the requirement, under an approved nondisruptive test plan.

1. Assume gateway state synchronization includes backend data recovery
2. The gateway fault tolerance does not establish tolerance of backend failure
3. Claim the whole application has no single dependency because gateways are paired
4. Limit failure analysis to routers while promising service-wide resilience

Correct answer: B. The gateway fault tolerance does not establish tolerance of backend failure

Resilience claims should identify their scope. The required task still depends on the backend and its own continuity/recovery design.

1. Assume duplicate switches automatically replicate every effective security boundary
2. Accept required connectivity even when failover exposes prohibited administration
3. Check only normal-path firewall text and ignore alternate forwarding
4. Equivalent intended boundary enforcement on every usable path

Correct answer: D. Equivalent intended boundary enforcement on every usable path

Reliability changes must preserve security requirements. Trace alternate forwarding and verify both required connectivity and prohibited access during failure.

1. No; runtime at the new supported load must be established
2. Count UPS outlets rather than the supported load and required runtime
3. Reuse the 300-W runtime as proof of identical runtime at 500 W
4. Assume only network traffic rate affects battery duration

Correct answer: A. No; runtime at the new supported load must be established

Load and battery condition affect runtime. Use appropriate manufacturer/measurement evidence and the complete protected dependency chain under the actual design.

1. Record only redundancy worked without the injected failure or workload
2. Omit prohibited-access and normal-restoration verification
3. The exact failure case, affected paths, task recovery, capacity/security results, and restoration steps
4. Use a configuration screenshot as proof of all recovery durations

Correct answer: C. The exact failure case, affected paths, task recovery, capacity/security results, and restoration steps

A reviewable result needs a defined experiment. Otherwise later readers cannot know which resilience claim was tested or whether normal operation was restored safely.

1. Accept the alternate path from small ping success alone
2. Assume the primary path's MTU and control handling apply unchanged
3. Treat route presence as proof that all required payload sizes transfer
4. Payload-size behavior and required control feedback on the actual backup path

Correct answer: D. Payload-size behavior and required control feedback on the actual backup path

The alternate path has different constraints. Test representative transfers and relevant MTU/error behavior in addition to small reachability probes.

1. Assume upstream resilience removes endpoint cable and NIC dependencies
2. Count the second switch as automatic alternate attachment for every singly connected endpoint
3. Claim DNS can move a physical Ethernet attachment during switch loss
4. Endpoint attachment can still be a single dependency for that endpoint

Correct answer: D. Endpoint attachment can still be a single dependency for that endpoint

Upstream resilience does not fix a sole endpoint attachment. State the required endpoint failure coverage and supported alternate design rather than count unused nearby devices.

1. Restore routes and omit management identity validation
2. Place all private credentials in a public recovery document for convenience
3. Secure credential/certificate recovery and identity dependencies under approved handling
4. Assume saved configuration text contains every required certificate and protected key

Correct answer: C. Secure credential/certificate recovery and identity dependencies under approved handling

Configuration text may not contain all required secrets or identity state. Plan protected recovery, access, renewal, and verification without exposing credentials.

1. Treat one broadcast domain as proof of secure communication among every role
2. Choose the largest flat subnet solely because it has the fewest settings
3. Broadcast/failure scope, security boundaries, capacity, and maintainability against requirements
4. Ignore broader shared impact because initial configuration is shorter

Correct answer: C. Broadcast/failure scope, security boundaries, capacity, and maintainability against requirements

A flat design may simplify some tasks while expanding shared impact and weakening intended boundaries. Evaluate concrete workload and access needs rather than assume fewer settings means better design.

1. Assume redundant routers duplicate every DHCP dependency automatically
2. Exclude address-service lifecycle from the required task's availability model
3. Use existing clients' success as proof that new clients can obtain configuration
4. Established operation versus onboarding and lease lifecycle dependencies

Correct answer: D. Established operation versus onboarding and lease lifecycle dependencies

Failure impact can depend on client state and time. Test new admission, renewal/rebinding, and recovery under the relevant DHCP design rather than only established traffic.

1. Verify only reachability and ignore where session state is held
2. Forward/reverse path selection, state placement, and supported symmetry/state-sharing behavior
3. Assume any forward and reverse routes are interchangeable across stateful devices
4. Widen all policies instead of examining the asymmetric state path

Correct answer: B. Forward/reverse path selection, state placement, and supported symmetry/state-sharing behavior

Path redundancy can interact with stateful devices. Verify the actual session context and supported architecture instead of assuming any pair of routes is interchangeable.

1. Omit remaining shared dependencies to make the redundancy claim simpler
2. Present only a device-rich diagram and claim complete resilience
3. A requirement-to-path map with capacities, failure domains, security boundaries, recovery tests, and remaining limits
4. Choose brands before defining task capacity and failure requirements

Correct answer: C. A requirement-to-path map with capacities, failure domains, security boundaries, recovery tests, and remaining limits

A design is reviewable when its choices and limits are connected to required behavior. Testable assumptions help reviewers identify coverage and unresolved dependencies before deployment.