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.
Your organization is migrating a latency-sensitive HTTP microservice running in a routes-based GKE Standard cluster to Google Cloud. Currently, the cluster exposes the application through an unmanaged instance group behind an external Application Load Balancer, and you need to transition to container-native load balancing with the following requirements:
kube-proxy and extra node-to-node network hops.Which architecture and configuration should you implement?
GCE_VM_IP_PORT NEGs for the GKE Service and GCE_VM_IP NEGs for the legacy VMs to the same backend service, and rely on Pod readinessProbe initialDelaySeconds instead of a Compute Engine health check so the NEG controller manages Pod readiness gates automatically.This configuration attempts to combine GCE_VM_IP_PORT NEGs (which contain IP:port endpoints for Pods) and GCE_VM_IP NEGs (which contain primary VM interface IPs without ports) on one backend service while delegating Pod readiness solely to Kubernetes readinessProbe settings.
GCE_VM_IP or all GCE_VM_IP_PORT. Additionally, GCE_VM_IP NEGs are only supported by Passthrough Network Load Balancers, not Application Load Balancers.cloud.google.com/load-balancer-neg-ready) to True as soon as the endpoint is programmed in the NEG. This can cause the load balancer to send traffic to Pods that are not yet healthy from the load balancer's perspective.initialDelaySeconds misconception: The initialDelaySeconds parameter inside a Pod's readinessProbe is enforced only by kubelet inside the node, not by the external Compute Engine container-native load balancer health check.GCE_VM_IP_PORT zonal NEGs, which can reference either Pod alias IPs or VM primary IPs with ports.130.211.0.0/22 and 35.191.0.0/16, backend endpoints cannot be verified by Google Cloud health probers.Both the GKE Pods and legacy Compute Engine VMs were placed in GCE_VM_IP_PORT zonal NEGs and an explicit Compute Engine health check with appropriate firewall rules was configured on the backend service.
LoadBalancer Service with GKE subsetting enabled (GCE_VM_IP NEGs) and externalTrafficPolicy: Local, and configure a health check on TCP port 10256 to validate Pod readiness directly.A Kubernetes Service of type: LoadBalancer provisions a Layer 4 Passthrough Network Load Balancer rather than an Application Load Balancer. Enabling GKE subsetting groups cluster node VMs into zonal GCE_VM_IP NEGs so that only nodes hosting serving Pods (or eligible subsetted nodes) are targeted by the load balancer.
GCE_VM_IP NEGs target the primary internal IP address of node VMs, not Pod alias IP addresses. Traffic still lands on the node VM and relies on kube-proxy or eBPF (cilium-agent) on the node to reach the Pod.LoadBalancer Services are answered by kube-proxy or cilium-agent on each node, never directly by Pods.Local policy: TCP port 10256 is used when externalTrafficPolicy is set to Cluster. When externalTrafficPolicy is Local, the control plane assigns a health check port from the node port range.GCE_VM_IP), it cannot report per-Pod latency metrics.LoadBalancer Services do not use GCE_VM_IP_PORT container-native load balancing.GCE_VM_IP_PORT container-native load balancing.You needed a Layer 4 Passthrough Network Load Balancer for non-HTTP TCP/UDP traffic targeting GKE nodes while improving internal load balancer node scalability beyond 250 nodes.
Even with externalTrafficPolicy: Local reducing cross-node hops, node-level GCE_VM_IP NEGs cannot provide Pod-level latency visibility or direct Pod health checking via GCE_VM_IP_PORT endpoints.
cloud.google.com/neg: '{"ingress": true}' annotation on the Service so GKE provisions the load balancer and health checks, and attach the legacy Compute Engine VM instance group to the Ingress-managed backend service.When you deploy an Ingress resource in a VPC-native GKE cluster with the cloud.google.com/neg: '{"ingress": true}' annotation on the referenced Service, the GKE Ingress controller automatically provisions and manages the entire external Application Load Balancer stack. This includes the forwarding rule, virtual IP address, URL map, backend services, Compute Engine health checks, ingress firewall rules, and zonal GCE_VM_IP_PORT NEGs.
GCE_VM_IP_PORT) as backend groups; all backend groups on a backend service must be of the same type.GCE_VM_IP_PORT NEGs with external Ingress and instead fall back to unmanaged instance groups.The migration only involved workloads already running inside GKE clusters and there was no requirement to share the same backend service VIP with legacy Compute Engine VMs.
To serve both GKE Pods and Compute Engine VMs behind a single Application Load Balancer VIP, you must use standalone zonal GCE_VM_IP_PORT NEGs and manage the backend service outside of the GKE Ingress controller.
GCE_VM_IP_PORT NEGs, attach both the GKE NEGs and GCE_VM_IP_PORT NEGs for the legacy VMs to the Application Load Balancer backend service, and create a Compute Engine health check with ingress firewall rules allowing 130.211.0.0/22 and 35.191.0.0/16 on the Pod target port.GCE_VM_IP_PORT NEGs in VPC-native clusters areContainer-native load balancing uses zonal Network Endpoint Groups (NEGs) of type GCE_VM_IP_PORT so that Google Cloud proxy-based load balancers distribute traffic directly to individual Pod IP addresses and container ports (IP:targetPort). In a VPC-native cluster, Pod IP addresses are allocated from VPC alias IP ranges (a secondary subnet range for Pods), making every Pod IP natively routable within the VPC without Network Address Translation (NAT). Unlike Ingress-managed NEGs, standalone NEGs are populated by the GKE NEG controller while allowing you to manually configure and attach them to self-managed backend services.
kube-proxy and extra hops: By targeting Pod alias IP addresses in GCE_VM_IP_PORT NEGs, the external Application Load Balancer sends requests straight to the destination Pod, eliminating the node-level NodePort hop and iptables translation by kube-proxy.GCE_VM_IP_PORT NEGs allow a single backend service VIP to combine GKE Pod endpoints with GCE_VM_IP_PORT NEGs containing legacy Compute Engine VM primary IPs and ports.130.211.0.0/22 and 35.191.0.0/16) on the Pod's target port allows the GKE NEG controller to manage Pod readiness gates accurately, transitioning Pods to Ready only after load balancer probes pass.ServiceNetworkEndpointGroup (svcneg) custom resources.While GKE Ingress automates load balancer creation, it is opinionated and does not permit attaching external VM backends alongside GKE Pods on the same backend service. Standalone zonal GCE_VM_IP_PORT NEGs provide full container-native load balancing benefits alongside the architectural flexibility required for heterogeneous migrations, provided you manage the Compute Engine health checks, firewall rules, and NEG lifecycle cleanup.
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