Explore / Google Cloud Professional Cloud Network Engineer Exam Prep / Practice questions
Google Cloud Professional Cloud Network Engineer practice questions
10 questions from the bank behind this course's timed mock. Free to read and to answer, no account needed, and each has its answer and an explanation. They belong to Google Cloud Professional Cloud Network Engineer Exam Prep.
Answer them one at a time →Read lesson 1 free
Question 1
A retailer is choosing how to connect its data centre to Google Cloud. The scenario requires a 99.99% availability SLA with no single point of failure. The last sentence asks for the most cost-effective option that meets these requirements. Using the scenario-question method, how should you weigh the requirements?
- Give cost and availability equal weight and pick the option with the best overall balance between them.
- Treat operations effort as the binding constraint, because a managed service is always the preferred answer.
- Treat cost as the binding constraint. Pick the cheapest option, then check whether its availability is close enough to 99.99%.
- Treat the 99.99% SLA as the binding constraint. Rule out every option that can't meet it, then pick the cheapest of the options left.
Show the answer
Answer: D. Treat the 99.99% SLA as the binding constraint. Rule out every option that can't meet it, then pick the cheapest of the options left.
The 99.99% SLA can't be traded away, so it decides which options survive. 'Most cost-effective' only chooses between options that already meet it. Starting from the cheapest option gets the order backwards and can land you on an answer that breaks the SLA.
Question 2
A network engineer needs full private routing between two VPCs owned by different teams. vpc-red has a subnet 10.20.0.0/16 and vpc-blue has a subnet 10.20.8.0/22. Workloads on each side need to reach many hosts on the other side, not just one service. What should the engineer do so that VPC Network Peering can be used?
- Re-address one of the subnets so that no range in one VPC matches, contains or fits inside a range in the other, then create the peering from both sides
- Publish vpc-blue's workloads through a Private Service Connect service attachment and create an endpoint in vpc-red
- Create the peering with --export-custom-routes and --import-custom-routes so the more specific /22 route takes priority
- Create the peering configuration only in vpc-red's project so that routes flow one way and avoid the conflict
Show the answer
Answer: A. Re-address one of the subnets so that no range in one VPC matches, contains or fits inside a range in the other, then create the peering from both sides
Peering doesn't allow any subnet in one network to match, contain or fit inside a subnet in the peer, and 10.20.8.0/22 sits inside 10.20.0.0/16, so one side has to be renumbered first. PSC does work with overlapping IPs, but it exposes a single service behind a load balancer and doesn't give the many-host routing these teams need.
Question 3
You are a network engineer at a retailer. A latency-sensitive Cloud Run service must call internal APIs on VMs in prod-vpc. The platform team wants the lowest latency and highest throughput, and does not want to pay for connector instances. The target subnet has plenty of spare IP addresses. What should you configure?
- A private services access allocated range peered with Service Networking
- Direct VPC egress for the Cloud Run service
- A Private Service Connect endpoint using the all-apis bundle
- A Serverless VPC Access connector in a dedicated /28 subnet
Show the answer
Answer: B. Direct VPC egress for the Cloud Run service
Direct VPC egress gives Cloud Run lower latency and higher throughput with no connector VMs to pay for. Its cost is higher IP usage, and this subnet has addresses to spare. A Serverless VPC Access connector would also reach the VMs, but it adds connector instances and their cost, which the team wants to avoid.
Question 4
A data engineering team runs a Dataflow job in a project inside an enforced service perimeter that restricts storage.googleapis.com. The job must now write daily export files to a Cloud Storage bucket owned by a partner, in a project outside the perimeter. All other data movement out of the perimeter must stay blocked. What should the network engineer configure?
- An access level containing the partner's public IP ranges, added to the perimeter
- An egress rule on the perimeter that allows the Dataflow job's service account to use storage.googleapis.com against the partner's project
- An ingress rule on the perimeter that names the partner's project as the source and storage.googleapis.com as the service
- A VPC firewall egress rule that allows the Dataflow workers to reach the partner bucket's endpoint
Show the answer
Answer: B. An egress rule on the perimeter that allows the Dataflow job's service account to use storage.googleapis.com against the partner's project
Here a workload inside the perimeter is reaching out to a resource outside it, so you need an egress rule (egressFrom/egressTo) that pins the identity, the service and the target project. An ingress rule covers the opposite direction, a call from outside coming in, so it would not allow this write.
Question 5
Your organization uses Shared VPC. A platform team in a service project creates a GKE cluster with private nodes on a subnet owned by the host project. Security policy forbids any outbound internet path, so no Cloud NAT gateway exists. Pods fail to pull images from Artifact Registry. What should you do to fix this while keeping to the policy?
- Add the node subnet range to the cluster's authorized networks
- Ask the host project administrator to enable Private Google Access on the cluster's node subnet
- Enable the DNS-based control plane endpoint with --enable-dns-access
- Create a Cloud Router and a Cloud NAT gateway covering the node and Pod ranges
Show the answer
Answer: B. Ask the host project administrator to enable Private Google Access on the cluster's node subnet
GKE turns on Private Google Access automatically except in Shared VPC, where the host project admin has to enable it so that private nodes can reach Google APIs such as Artifact Registry. Cloud NAT would also give a route out, but it opens internet egress, which the policy forbids.
Question 6
Your company serves shop.example.com through a classic Application Load Balancer. The release team wants to send 5% of requests to a new backend service and needs to be able to roll back within seconds. The change should stay inside Google Cloud load balancing configuration. What should you recommend?
- Create a second load balancer for the new version and use a Cloud DNS weighted round robin routing policy to send 5% of lookups to it
- Export the classic load balancer's URL map, add weightedBackendServices to the path matcher's routeRules, and import it again
- Migrate to the global external Application Load Balancer, then add a route rule that uses weightedBackendServices with weights such as 95 and 5
- Add a requestMirrorPolicy with mirrorPercent set to 5.0 that points to the new backend service
Show the answer
Answer: C. Migrate to the global external Application Load Balancer, then add a route rule that uses weightedBackendServices with weights such as 95 and 5
Weighted backend splitting is only available on the Envoy-based load balancers, so you have to move off the classic Application Load Balancer before you can set up a per-request canary that you can roll back by editing the weights. DNS weighting is a real feature, but resolvers cache records until the TTL expires, so rollback isn't instant and the split isn't per request. Mirroring throws away the shadow's responses, so no user would ever be served by the new version.
Question 7
A network engineer at a retailer orders a layer 3 Partner Interconnect service in europe-west2. The provider's routers will exchange routes with Google. The engineer has created the Cloud Router (ASN 16550) and the Partner VLAN attachment, and the provider has finished its side. Once the attachment is activated, what BGP configuration does the engineer still need to do on the Cloud Router?
- Run gcloud compute routers update-bgp-peer to set the peer ASN to the on-premises router's ASN.
- None. Google adds the provider's ASN to the BGP session automatically, and the provider sets up BGP with the Cloud Router.
- Choose custom link-local IP ranges for the BGP session, then create the BGP peer by hand.
- Add a BGP peer with an MD5 authentication key that matches the on-premises router's key.
Show the answer
Answer: B. None. Google adds the provider's ASN to the BGP session automatically, and the provider sets up BGP with the Cloud Router.
On a layer 3 service the provider runs BGP with your Cloud Router, and Google adds the provider's ASN for you, so the Cloud Router needs no BGP work from you. Setting the peer ASN to your on-premises ASN is the extra step for a layer 2 service, where your own router is the BGP peer.
Question 8
A network engineer connects an on-premises data centre to prod-vpc over HA VPN. Cloud Router cr-onprem-use1 uses the default advertisement mode. Security policy says on-premises hosts must reach Google APIs only through restricted.googleapis.com, and on-premises must keep receiving routes to every VPC subnet. Which TWO actions are required? (Select TWO.)
More than one answer is right.
- Create a custom static route in the VPC for 199.36.153.4/30 with the default internet gateway as its next hop
- Set the Cloud Router to --advertisement-mode=custom with --set-advertisement-groups=all_subnets and --set-advertisement-ranges=199.36.153.4/30
- Switch prod-vpc to --bgp-best-path-selection-mode=STANDARD so the VIP range is compared by AS path
- Add 199.36.153.4/30 as a custom learned route range on the on-premises BGP session
- Set the Cloud Router to --advertisement-mode=custom with only --set-advertisement-ranges=199.36.153.4/30
Show the answer
Answers: A. Create a custom static route in the VPC for 199.36.153.4/30 with the default internet gateway as its next hop; B. Set the Cloud Router to --advertisement-mode=custom with --set-advertisement-groups=all_subnets and --set-advertisement-ranges=199.36.153.4/30
Google doesn't publish routes to the restricted VIP. You advertise 199.36.153.4/30 to on-premises yourself (keeping all_subnets so the subnet routes stay), and you pair it with a VPC static route to the default internet gateway. A custom learned route works in the wrong direction: it installs a route in the VPC as if on-premises had advertised the prefix, so on-premises still never learns the VIP.
Question 9
You are the network engineer for a retail site behind a global external Application Load Balancer. The 5xx ratio for backend service 'orders-bes' has gone from 0.1% to 8%. No Cloud Armor policy changed today. In the load balancer logs, most of the 5xx entries show statusDetails="failed_to_connect_to_backend". What is the most likely cause, and where should you look first?
- A Cloud Armor rule with action deny-502 is blocking requests. Check the security policy's enforced rules.
- Clients on slow networks are timing out. Compare p99 total_latencies with frontend_tcp_rtt.
- The load balancer proxy cannot reach the backends. Check the firewall rules, backend health and backend capacity for orders-bes.
- The application code in orders-bes is returning errors. Check the application logs for stack traces.
Show the answer
Answer: C. The load balancer proxy cannot reach the backends. Check the firewall rules, backend health and backend capacity for orders-bes.
failed_to_connect_to_backend means the proxy produced the 5xx itself because it could not connect, which usually points to health, firewall or capacity problems. If the application code had returned the error, the logs would show response_sent_by_backend, and a Cloud Armor block would show denied_by_security_policy.
Question 10
A retail company runs its checkout service as a regional managed instance group spread across europe-west2-a, europe-west2-b and europe-west2-c in one VPC network. The security team must turn on Cloud NGFW Enterprise intrusion prevention, including decrypting HTTPS, for all instances. They want the fewest resources that give full coverage. Which set of resources should they create?
- One firewall endpoint in europe-west2-a, one endpoint association, and custom static routes that send traffic from europe-west2-b and europe-west2-c through that endpoint
- One firewall endpoint in europe-west2, one endpoint association with the VPC network, and one Certificate Authority Service CA pool with one TLS inspection policy in europe-west2
- Three zonal firewall endpoints, three endpoint associations, and a separate CA pool and TLS inspection policy for each of the three zones
- Three zonal firewall endpoints (one in each zone), three endpoint associations with the VPC network, and one Certificate Authority Service CA pool with one TLS inspection policy in europe-west2
Show the answer
Answer: D. Three zonal firewall endpoints (one in each zone), three endpoint associations with the VPC network, and one Certificate Authority Service CA pool with one TLS inspection policy in europe-west2
Firewall endpoints are zonal and only inspect workloads in their own zone, so each zone needs its own endpoint and association. The CA pool and TLS inspection policy are regional, so one of each covers europe-west2. A per-zone CA pool and TLS policy adds resources that do nothing, and routes can't send traffic to an endpoint because redirection is transparent and zone-bound.
The full timed mock
50 questions in 120 minutes, with a pass mark of 70%. Each paper is drawn from a bank of questions written for the mock alone, and these 10 are left out of it. Full access opens it, with every level of the course.
LevelUpWith is independent of the organisation that sets this exam. These questions were written for this course; they are not taken from the exam.