350-401

Free 350-401 practice questions

10 questions from our 350-401 practice tests, each with the correct answer and a written explanation. Read them with the answers shown, or switch to quiz mode and answer first.

Back to the 350-401 practice exams

Question 1 of 10, Architecture

A hospital using SD-Access must completely isolate medical devices from guest users, and within the corporate network it must prevent HR laptops from talking to engineering laptops while both share a subnet. Which design meets both requirements?

  1. ASeparate virtual networks for medical, guest and corporate, plus SGT policies inside the corporate VN Correct answer
  2. BSeparate VRFs for HR and engineering, plus one shared VN for medical devices and guests
  3. CSGT policies for medical and guest isolation, and separate subnets for HR and engineering
  4. DOne virtual network for all users, with VLAN ACLs between the HR and engineering VLANs

Answer and explanation: Virtual networks (VRFs) provide macro-segmentation with fully separate routing for medical, guest and corporate traffic, while scalable group tags provide micro-segmentation between groups in the same VN and subnet. A single VN cannot fully isolate the groups, separate VRFs for HR and engineering contradict the shared-subnet requirement, and putting medical devices and guests in one VN relies on policy instead of routing isolation.

Question 2 of 10, Architecture

A regional hospital campus has three buildings, about 600 users and a single server room. The network team wants to reduce the number of switching layers while keeping a redundant design that can grow modestly. Which design is the MOST appropriate?

  1. AA flat Layer 2 design that extends every VLAN between all buildings with Rapid PVST+
  2. BA two-tier collapsed-core design in which the distribution switches also perform the core role Correct answer
  3. CA leaf-spine fabric in which every access switch connects to every spine switch
  4. DA three-tier design with separate access, distribution and core layers in every building

Answer and explanation: A collapsed core merges the core and distribution layers, which suits a small or medium campus with few distribution blocks while keeping redundant uplinks. A dedicated three-tier design adds cost and complexity that only pays off with many distribution blocks. Leaf-spine is a data center fabric design, and stretching every VLAN across buildings enlarges failure domains and relies on spanning tree for redundancy.

Question 3 of 10, Virtualization

A data center team needs a Layer 2 overlay that can support more than 4094 isolated segments across a routed IP underlay. Which technology is designed for this, and how large is its segment identifier?

  1. ALISP, which uses a 16-bit instance ID for each virtual network
  2. B802.1ad Q-in-Q, which stacks two 12-bit VLAN tags in each frame
  3. CGRE, which carries a 32-bit key field to identify each segment
  4. DVXLAN, which carries a 24-bit VXLAN network identifier in UDP Correct answer

Answer and explanation: VXLAN encapsulates Ethernet frames in UDP (port 4789) with a 24-bit VNI, allowing about 16 million segments over a routed underlay. Q-in-Q extends VLAN tagging but is a Layer 2 technique, GRE keys are not a standard segmentation scheme for overlays, and the LISP instance ID is 24 bits and LISP itself is the control plane rather than the Layer 2 overlay.

Question 4 of 10, Infrastructure

Which statement correctly describes the difference between NTP and the Precision Time Protocol (PTP, IEEE 1588)?

  1. APTP synchronizes clocks only across WAN links, while NTP is limited to devices within a single Layer 2 domain
  2. BNTP and PTP use the same stratum hierarchy, but PTP encrypts timestamps while NTP sends them in clear text
  3. CPTP is designed for sub-microsecond accuracy with hardware timestamping, while NTP typically provides millisecond-level accuracy Correct answer
  4. DNTP uses hardware timestamping and grandmaster clocks to reach nanosecond accuracy, while PTP relies on software stratum levels

Answer and explanation: PTP uses hardware timestamping, a best master clock algorithm and boundary or transparent clocks to deliver sub-microsecond precision for industrial, financial and media applications. NTP uses a stratum hierarchy and typically reaches millisecond accuracy. The first option swaps the two protocols, PTP does not reuse NTP stratum levels, and neither protocol is restricted to WAN-only or single-LAN use.

Question 5 of 10, Infrastructure

Two routers peer with eBGP using their loopback addresses, and the session stays in the Active state. Both routers have static routes to the peer loopback and 'neighbor x.x.x.x update-source Loopback0'. What is the MOST likely missing configuration?

  1. Aneighbor x.x.x.x soft-reconfiguration inbound
  2. Bneighbor x.x.x.x remote-private-as
  3. Cneighbor x.x.x.x ebgp-multihop 2 Correct answer
  4. Dneighbor x.x.x.x next-hop-self

Answer and explanation: eBGP packets are sent with a TTL of 1 by default, so a session between loopbacks needs 'ebgp-multihop' (or disable-connected-check on a directly connected link) to establish. Next-hop-self changes advertised next hops, remove-private-as affects AS paths in updates, and soft reconfiguration stores received updates; none of them helps the TCP session come up.

Question 6 of 10, Infrastructure

An EIGRP router has a successor to 172.16.5.0/24 with feasible distance 3000. Neighbor X reports the same prefix with a reported distance of 2800 and a total metric of 3600. Neighbor Y reports a reported distance of 3100 and a total metric of 3300. Which neighbor is a feasible successor?

  1. ANeighbor X only, because its reported distance is lower than the feasible distance Correct answer
  2. BNeighbor Y only, because its total metric is closer to the current feasible distance
  3. CBoth neighbors, because each has a total metric within twice the feasible distance
  4. DNeither neighbor, because both total metrics are higher than the feasible distance

Answer and explanation: The feasibility condition requires the reported distance of an alternate path to be strictly lower than the current feasible distance, so only Neighbor X (2800 < 3000) qualifies. Neighbor Y's reported distance (3100) fails the condition regardless of its total metric. There is no rule based on twice the feasible distance, and a feasible successor's total metric is normally higher than the successor's.

Question 7 of 10, Network Assurance

An engineer runs a traceroute from a branch router and sees the first three hops reply normally, followed by lines of asterisks until the trace ends, while users can still reach the destination application. What is the MOST likely explanation?

  1. AThe destination is unreachable because the fourth hop has no route to it
  2. BThe third hop is dropping all of the traffic because of an MTU mismatch on its outgoing link
  3. CThe traceroute TTL value was set too low to reach the destination
  4. DDevices past the third hop filter or do not send ICMP replies Correct answer

Answer and explanation: Asterisks mean no ICMP reply was received, which commonly happens when firewalls or routers do not generate or permit ICMP time-exceeded or port-unreachable messages; since the application works, the path itself is fine. A routing failure would also break the application, a low maximum TTL would end the trace rather than produce timeouts, and an MTU problem affects large packets rather than traceroute probes.

Question 8 of 10, Security

An engineer applies Control Plane Policing to protect the route processor from excessive SSH and ICMP traffic. Which command attaches the policy correctly?

  1. Aline vty 0 15 followed by access-class COPP in
  2. Bcontrol-plane followed by service-policy output COPP
  3. Ccontrol-plane followed by service-policy input COPP Correct answer
  4. Dinterface GigabitEthernet0/0 followed by service-policy input COPP

Answer and explanation: CoPP is configured by entering 'control-plane' configuration mode and applying the policy map with 'service-policy input', which polices traffic destined to the route processor. Applying the policy to a physical interface affects transit traffic, the output direction controls traffic sent by the control plane, and a VTY access-class filters management sessions but does not police other protocols such as ICMP.

Question 9 of 10, Security

An ISP edge router must drop packets arriving on the customer interface whose source addresses are not reachable back through that same interface, to block spoofed traffic. Which feature should be enabled?

  1. ADynamic ARP inspection on the customer-facing VLAN
  2. BUnicast RPF in strict mode on the customer-facing interface Correct answer
  3. CAn outbound ACL on the customer interface denying private ranges
  4. DUnicast RPF in loose mode on the upstream provider interface

Answer and explanation: Strict uRPF checks that the source address is reachable via the receiving interface and drops packets that fail the check, which blocks spoofed sources from a single-homed customer. Loose mode only checks that a route exists on any interface, an outbound ACL filters traffic toward the customer rather than from it, and DAI validates ARP packets on a Layer 2 network.

Question 10 of 10, Automation and Artificial Intelligence

An anomaly detection model was trained on a campus's traffic patterns a year ago. After a move to hybrid work and new SaaS applications, it raises many false alarms. What is the MOST likely cause and remedy?

  1. AA hardware fault in the flow collectors; replace them, because trained models cannot produce false alarms
  2. BA licensing limit; reduce the number of monitored devices until alarms stop
  3. CData drift; retrain or recalibrate the model on recent telemetry that reflects the new patterns Correct answer
  4. DOverfitting to new data; restore the year-old model and disable automatic updates

Answer and explanation: When the underlying traffic patterns change, the model's learned baseline no longer matches reality (data drift), so it must be retrained or recalibrated on current data. Restoring an outdated model keeps the problem, models do produce false positives, and reducing monitored devices hides symptoms instead of fixing them.