PCNE

Free PCNE practice questions

10 questions from our PCNE 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 PCNE practice exams

Question 1 of 10, Designing and planning a Google Cloud VPC network

A company has an on-premises Active Directory DNS that is authoritative for corp.example.com and a Shared VPC in Google Cloud whose private zone gcp.example.com must be resolvable from on-premises. Workloads in Google Cloud must also resolve corp.example.com. The network team wants a single, centrally managed configuration in the host project. Which hybrid DNS design meets these requirements?

  1. AA Cloud DNS private zone for corp.example.com that copies all AD records, plus a public zone for gcp.example.com that on-premises resolvers query over the internet
  2. BA Cloud DNS peering zone for corp.example.com that targets the on-premises network over Interconnect, plus an outbound server policy that forwards gcp.example.com queries arriving from on-premises resolvers
  3. CA Cloud DNS forwarding zone for corp.example.com that targets the AD DNS servers, plus an inbound server policy so on-premises DNS can conditionally forward gcp.example.com to the inbound entry points Correct answer
  4. DA forwarding zone for corp.example.com in every service project, plus Cloud DNS response policies that rewrite gcp.example.com answers for on-premises clients

Answer and explanation: A forwarding zone sends queries for corp.example.com to the on-premises authoritative servers, and an inbound server policy creates entry-point addresses in the VPC that on-premises resolvers can forward gcp.example.com to; both live in the host project. DNS peering zones target other VPC networks, not on-premises networks, and outbound server policies redirect all VPC queries rather than receiving on-premises ones. Copying AD records into a private zone and exposing internal names publicly creates drift and leaks internal names. Per-project forwarding zones break central management, and response policies only affect resolution inside VPCs.

Question 2 of 10, Designing and planning a Google Cloud VPC network

An analytics team in a VPC needs private connectivity to a Cloud SQL instance, an AlloyDB cluster, and a third-party SaaS database published by a partner as a Private Service Connect service attachment. Address space in the consumer VPC is scarce, and the security team wants each managed service to consume only a single internal IP address in the consumer network where possible. Which connectivity plan fits best?

  1. AUse Private Service Connect endpoints for Cloud SQL, AlloyDB, and the partner service so that each consumes one consumer IP address Correct answer
  2. BUse private services access for all three services by asking the partner to join the servicenetworking peering connection with its own range
  3. CUse Serverless VPC Access connectors for Cloud SQL and AlloyDB, and VPC Network Peering directly to the partner's producer VPC network
  4. DUse private services access for Cloud SQL and AlloyDB with a shared allocated /16 range, and a Private Service Connect endpoint only for the partner service

Answer and explanation: Cloud SQL and AlloyDB both support Private Service Connect, and a PSC endpoint consumes a single IP address in the consumer subnet, which satisfies the address scarcity requirement; the partner service is already published through a service attachment. Private services access reserves an allocated range in the consumer VPC, usually much larger than one address per service. Serverless VPC Access connectors are for serverless workloads reaching a VPC, not for VMs reaching managed databases. Third parties cannot join the Google-managed servicenetworking peering, and direct peering to a partner VPC exchanges subnet routes broadly.

Question 3 of 10, Implementing a VPC network

A team wants traffic from subnet 10.30.0.0/24 destined for 10.40.0.0/16 to be sent through an internal passthrough Network Load Balancer fronting inspection appliances, while traffic from other subnets to 10.40.0.0/16 follows normal subnet routes. Destination subnet routes cannot be overridden by static routes. What should they configure?

  1. AA static route to 10.40.0.0/16 with the internal load balancer as next hop and the network tag of the VMs in 10.30.0.0/24
  2. BA Cloud Router custom advertisement for 10.40.0.0/16 with a lower MED pointing to the inspection appliances
  3. CA static route to 10.40.0.0/16 with the internal load balancer as next hop and a priority of 0 so that it always takes precedence over the subnet routes
  4. DA policy-based route that matches source 10.30.0.0/24 and destination 10.40.0.0/16 with the internal load balancer as next hop Correct answer

Answer and explanation: Policy-based routes can match on source range and are evaluated before other route types, so they can steer traffic from one subnet through an internal passthrough load balancer even toward destinations covered by subnet routes. Static routes, even with the best priority, cannot override subnet routes with equal or more specific destinations. Adding a network tag only limits which VMs use a static route; it does not let the route beat the more specific subnet routes. Cloud Router advertisements affect hybrid peers, not routing inside the VPC.

Question 4 of 10, Implementing a VPC network

A GKE cluster sends traffic from Pods to an on-premises service that only accepts connections from node IP addresses. Pods use a non-RFC 1918 secondary range. What should the engineer configure so on-premises sees node IPs?

  1. AEnable Dataplane V2 on the cluster so that Pod traffic is source-NATed automatically
  2. BEnable Private NAT on the cluster's subnet for traffic destined to the on-premises ranges
  3. CAdd the on-premises ranges to the nonMasqueradeCIDRs list of the ip-masq-agent
  4. DConfigure the ip-masq-agent with nonMasqueradeCIDRs that exclude the on-premises ranges Correct answer

Answer and explanation: The ip-masq-agent masquerades Pod traffic to the node IP for any destination not listed in nonMasqueradeCIDRs, so leaving the on-premises ranges out of that list causes SNAT to node IPs. Adding the on-premises ranges to nonMasqueradeCIDRs does the opposite and preserves Pod IPs. Private NAT is for overlapping networks across NCC spokes and hybrid connections and would translate to a NAT range, not node IPs. Dataplane V2 enforces network policy and does not change masquerade behavior on its own.

Question 5 of 10, Configuring managed network services

A shopping cart service behind an external Application Load Balancer stores session state in memory on each VM. During scale-out, some users lose their carts. The team needs requests from the same browser to reach the same backend, and many users share corporate NAT IP addresses. Which session affinity setting is the best fit?

  1. AClient IP and protocol affinity on the backend
  2. BGenerated cookie affinity on the backend service Correct answer
  3. CHeader field affinity on a header the client sets randomly per request
  4. DClient IP affinity on the backend service

Answer and explanation: Generated cookie affinity issues a load balancer cookie per browser, so each user returns to the same backend even when many users share a NAT IP. Client IP based affinity would map all users behind one corporate NAT to the same backend, causing imbalance, and adding protocol does not help. A header that changes per request provides no stickiness. Affinity also weakens during scale changes, so moving state to a shared store is a longer-term improvement.

Question 6 of 10, Configuring managed network services

A team is rolling out version 2 of a checkout service behind a global external Application Load Balancer. They want 5% of live requests sent to v2, and they also want an exact copy of all v1 traffic delivered to a shadow backend for testing without affecting responses. What should they configure in the URL map?

  1. AA weightedBackendServices action with 95 and 5 weights, plus a backend bucket configured as the default service for mirrored traffic
  2. BA weightedBackendServices route action with 95 and 5 weights, plus a requestMirrorPolicy that targets the shadow backend service Correct answer
  3. CTwo path matchers with different hosts, one for v2, plus a urlRewrite action that copies requests to the shadow backend
  4. DA Cloud Armor policy that redirects 5% of requests to v2, plus Packet Mirroring on the instance group serving v1 traffic

Answer and explanation: Weighted backend services in a route action split traffic by percentage, and a request mirror policy sends copies of requests to another backend service while ignoring its responses. URL rewrites change the host or path of a request but do not duplicate it, and separate hosts would not split live traffic by percentage. A default backend bucket does not receive mirrored requests. Cloud Armor redirect actions are not percentage-based traffic splitting, and Packet Mirroring copies packets at the VM level rather than routing HTTP requests to a test backend.

Question 7 of 10, Configuring and implementing hybrid and multicloud network interconnectivity

An on-premises site uses two Dedicated Interconnect VLAN attachments in the same region, and the team wants all traffic from Google Cloud toward on-premises to use attachment A, with attachment B used only if A fails. How should this preference be achieved?

  1. ASet a lower advertised route priority on the Cloud Router for routes it advertises over attachment A's BGP session
  2. BHave the on-premises routers advertise their prefixes with a lower MED over attachment A's BGP session than over attachment B's Correct answer
  3. CSet a lower advertised route priority on the Cloud Router for routes it advertises over attachment B's BGP session
  4. DEnable Bidirectional Forwarding Detection only on attachment A so that its routes are preferred while the session is up

Answer and explanation: Traffic from Google Cloud toward on-premises follows routes that Cloud Router learns, so the preference must be expressed in the attributes on-premises advertises; a lower MED learned over attachment A makes those routes preferred, and attachment B takes over if A's session fails. The advertised route priority on Cloud Router is sent to on-premises and influences traffic toward Google Cloud, which is the opposite direction, regardless of which session it is set on. BFD only speeds failure detection and does not create a preference.

Question 8 of 10, Configuring and implementing hybrid and multicloud network interconnectivity

A company runs analytics on AWS and must connect a Google Cloud VPC to an AWS VPC with 10 Gbps of dedicated, private bandwidth and without managing physical colocation equipment. Which solution should be used?

  1. ACross-Cloud Interconnect to AWS with VLAN attachments to a Cloud Router Correct answer
  2. BDedicated Interconnect from a colocation facility where both clouds have a presence
  3. CHA VPN between Google Cloud and an AWS virtual private gateway over the internet
  4. DPartner Interconnect from a provider that also offers AWS Direct Connect connections

Answer and explanation: Cross-Cloud Interconnect provides dedicated physical connections between Google and another cloud provider, managed by Google, so the customer does not operate colocation equipment. Dedicated Interconnect would require the customer to deploy and manage routers in the colocation facility. HA VPN runs over the internet and per-tunnel throughput is far below a dedicated 10 Gbps link. Partner Interconnect involves a third-party provider and is not a direct dedicated cloud-to-cloud connection.

Question 9 of 10, Managing, monitoring, and troubleshooting network operations

Applications behind a Cloud NAT gateway intermittently fail to open new connections to a single external API, while other destinations work. Cloud NAT uses automatic IP allocation and static port allocation with the default minimum ports per VM. Which observation in Cloud Monitoring confirms port exhaustion?

  1. AThe nat/allocated_ports metric shows fewer ports allocated than the configured minimum per VM
  2. BThe firewall/dropped_bytes_count metric increases on the egress deny rules for the subnet
  3. CThe nat/dropped_sent_packets_count metric shows drops with reason OUT_OF_RESOURCES Correct answer
  4. DThe router/bgp/session_up metric shows that the Cloud Router used by Cloud NAT has flapped

Answer and explanation: Drops with reason OUT_OF_RESOURCES indicate that the VM has run out of NAT source ports or NAT IPs, which is classic port exhaustion when many connections go to the same destination IP and port. Allocated ports below the minimum would indicate a configuration issue, not exhaustion. Cloud NAT does not depend on BGP sessions, so Cloud Router BGP state is unrelated. Egress firewall drops would block all destinations covered by the rule, not only under connection load.

Question 10 of 10, Configuring, implementing, and managing a cloud network security solution

Security wants copies of all traffic entering a set of payment VMs sent to a self-managed IDS cluster, without changing routing or affecting production traffic. What should be configured?

  1. AVPC Flow Logs with metadata on the payment subnet exported to the IDS cluster through Pub/Sub
  2. BA Packet Mirroring policy for the payment VMs with an internal passthrough Network Load Balancer in front of the IDS collectors Correct answer
  3. CA Cloud NGFW Enterprise firewall endpoint with a security profile that forwards copies to the IDS
  4. DA policy-based route that sends the payment VMs' traffic through an internal load balancer fronting the IDS appliances

Answer and explanation: Packet Mirroring clones traffic from selected instances and delivers it to collectors behind an internal passthrough Network Load Balancer, which is out-of-band and does not change production routing. A policy-based route puts appliances in the traffic path, which is in-band and can affect production. Flow Logs contain flow metadata, not packet payloads, so an IDS cannot inspect them. Cloud NGFW Enterprise performs its own inspection and does not forward copies to self-managed IDS.