Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Professional Cloud Network Engineer practice bank has an additional 2,437 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
You are deploying a VPC-native GKE cluster in a Shared VPC service project connected to on-premises via Dedicated Interconnect. Only a single /24 RFC 1918 range remains for GKE nodes. You must meet these requirements:
/17 IP range for Pods while remaining compatible with Windows workloads and legacy on-premises hardware./23 public IPv4 prefix (198.51.100.0/23) via Bring Your Own IP (BYOIP) for GKE Ingress external Application Load Balancers.Which IP addressing strategy should you implement?
/24 RFC 1918 range to the GKE node subnet, configure a /17 Class E (240.0.0.0/4) range as the Pod secondary range without SNAT, and import 198.51.100.0/23 via BYOIP in the Shared VPC host project for GKE Ingress.This design correctly places BYOIP in the Shared VPC host project for GKE Ingress, but attempts to solve IPv4 exhaustion by allocating Class E (240.0.0.0/4) non-RFC 1918 addresses to GKE Pods without any Source Network Address Translation (SNAT).
240.0.0.0/4) provides over 268 million IP addresses and works across Google Cloud VPCs, it is not supported on Windows operating systems and is rejected by many legacy on-premises routers, switches, and firewalls that treat 240.0.0.0/4 as reserved or invalid.240.0.0.0/4 source IP, causing the firewalls to drop the packets.240.0.0.0/4 addresses are not routable on the public internet, outbound Pod connections without SNAT cannot receive return traffic unless translated to a valid node or NAT IP.100.64.0.0/10) is recommended before considering Class E.240.0.0.0/4 addresses without unsupported registry modifications.The environment ran exclusively on Linux workloads, all on-premises networking hardware supported Class E routes, and SNAT (such as the IP masquerade agent or Hybrid NAT) was enabled for traffic leaving the VPC.
While Class E is an effective tool for mitigating IPv4 exhaustion in large Linux-only GKE clusters, RFC 6598 or PUPI combined with SNAT provides far safer interoperability in hybrid Windows and legacy enterprise environments.
198.51.100.0/23 via BYOIP in the Shared VPC service project, assign a /24 subset as the GKE node subnet and the remaining /24 for GKE Ingress, and use a separate /17 PUPI range for Pods without configuring the IP masquerade agent.This approach attempts to import a Bring Your Own IP (BYOIP) prefix directly inside a Shared VPC service project, repurpose half of that imported prefix as an internal VPC subnet for GKE nodes, and run Privately Used Public IP (PUPI) Pods without Source Network Address Translation (SNAT).
There is no scenario where this combination works; BYOIP must be provisioned in the host project (or a standalone VPC project), never assigned to primary or secondary IPv4 subnets, and PUPI Pods must use the IP masquerade agent for outbound communication.
A common misconception is that because BYOIP addresses belong to the customer, they can be partitioned freely between internal VPC subnets and external load balancers. In practice, BYOIP prefixes are exclusively for external public advertisement, whereas PUPIs are strictly internal subnet ranges.
/24 RFC 1918 range to the GKE node subnet, use a /17 Privately Used Public IP (PUPI) range distinct from 198.51.100.0/23 for the Pod secondary range with the IP masquerade agent configured to SNAT outbound Pod traffic, and import 198.51.100.0/23 via BYOIP in the Shared VPC host project for GKE Ingress.This design combines Privately Used Public IP (PUPI) addresses with the IP masquerade agent for internal Google Kubernetes Engine (GKE) Pod scaling, while properly provisioning Bring Your Own IP (BYOIP) addresses in the Shared VPC host project for external Application Load Balancers.
/17 Pod range with broad compatibility: Using a non-Class-E PUPI range (or an RFC 6598 100.64.0.0/10 range) avoids RFC 1918 exhaustion without introducing Class E (240.0.0.0/4) compatibility failures on Windows systems or legacy on-premises routers and firewalls.198.51.100.0/23 into the Shared VPC host project allows service projects to consume those static external IPs for GKE Ingress external forwarding rules, while keeping the BYOIP prefix strictly separate from any VPC primary or secondary PUPI subnet ranges.roles/compute.publicIpAdmin) role.Unlike Class E space, PUPI ranges coupled with node-level SNAT work seamlessly across mixed Windows and Linux environments and satisfy strict on-premises RFC 1918 firewall policies, while respecting all Shared VPC and BYOIP placement restrictions.
This option proposes migrating the architecture to a single-stack IPv6 GKE cluster with GKE Ingress and a global IPv6 BYOIP public delegated prefix (PDP) to bypass IPv4 exhaustion entirely.
gke-l7-global-external-managed, gke-l7-regional-external-managed, or gke-l7-rilb), and Layer 4 BYOIPv6 load balancing requires reserving a regional BYOIPv6 address for a Service of type LoadBalancer./23 IPv4 prefix for external load balancers. Furthermore, IPv6-only Pods cannot communicate with IPv4-only internet endpoints without configuring DNS64 and NAT64 translation, nor can they satisfy legacy IPv4-only on-premises systems without Private Service Connect or NAT64./96 per node from a /64 subnet and /112 per Pod, eliminating alias IP ranges, while Services receive a /112 from 2600:2D00:0:4::0:0/64.EXTERNAL_IPV6_FORWARDING_RULE_CREATION for regional external passthrough Network Load Balancers, or EXTERNAL_IPV6_SUBNETWORK_CREATION exclusively for VM subnets).The organization wanted an IPv6-only application tier exposed via the Kubernetes Gateway API using regional BYOIPv6 addresses (or a dual-stack frontend), and configured DNS64/NAT64 for IPv4 destination reachability rather than relying on an IPv4 /23 BYOIP block and GKE Ingress.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated