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 manage a VPC Packet Mirroring policy that mirrors all ingress and egress IPv4 traffic from production Compute Engine VMs to a self-managed collector Managed Instance Group (MIG) behind an internal passthrough Network Load Balancer. During peak traffic, you observe the following in Cloud Monitoring:
/mirroring/dropped_packets_count, egress bandwidth at roughly 3x baseline, and /mirroring/mirrored_bytes_count slightly exceeding sent_bytes_count + received_bytes_count.RATE_LIMIT_DROP in instance_network_dropped_packets_count.Which actions should you take to diagnose the drops and remediate saturation without disrupting active production connections?
Temporarily disabling a Packet Mirroring policy halts traffic cloning across matched VM interfaces, backend subsetting restricts the number of backend instances receiving connections from an internal passthrough Network Load Balancer, and VPC Flow Logs record sampled Layer 4 network flow metadata across VPC subnets.
This approach is invalid for Packet Mirroring remediation because disabling policies, deploying a single collector, and enabling backend subsetting all break core Packet Mirroring design constraints.
Unlike inline appliances, out-of-band collectors can be scaled horizontally inside a MIG without taking the mirroring policy offline, and dropped mirrored traffic must be inspected via /mirroring/dropped_packets_count on the mirrored VMs.
mirrored_bytes_count discrepancy, and switch the load balancer's session affinity to CLIENT_IP.Collector destination metrics on an internal passthrough Network Load Balancer aggregate total mirrored traffic received, VPC MTU defines the maximum Layer 3 packet payload size across the network, and CLIENT_IP session affinity pins all flows from a given client IP to a single backend VM.
/mirroring/dropped_packets_count./mirroring/mirrored_bytes_count exceeds sent_bytes_count + received_bytes_count is expected because mirrored byte metrics include Ethernet headers, whereas standard VM byte counters do not. Lowering the VPC MTU to 1400 bytes does not remove Ethernet header overhead and instead reduces payload efficiency and increases packet-per-second overhead.NONE (5-tuple hash), not CLIENT_IP, to distribute flows evenly across collectors./mirroring/dropped_packets_count.RATE_LIMIT_DROP) and slowing combined packet processing rates.NONE session affinity ensures individual flows from high-volume mirrored VMs spread across multiple collector backends rather than saturating one collector.Lowering MTU is occasionally used when encapsulating traffic (such as GENEVE or IPsec tunnels) causes path fragmentation, rather than to reconcile expected header accounting differences in Cloud Monitoring.
Understanding the exact header accounting of Compute Engine network metrics prevents unnecessary MTU changes that could destabilize production application traffic.
/mirroring/dropped_packets_count on individual mirrored VMs and instance_network_dropped_packets_count on collectors, update the active policy's filters to narrow mirrored traffic without disabling the policy, and scale out the collector MIG behind the load balancer with NONE session affinity and backend subsetting disabled.This approach pairs Cloud Monitoring telemetry from both mirrored source VMs and collector VMs with live Packet Mirroring policy filter updates and horizontal Managed Instance Group (MIG) scaling behind a properly configured internal passthrough Network Load Balancer.
/mirroring/dropped_packets_count, /mirroring/mirrored_packets_count, and /mirroring/mirrored_bytes_count. Meanwhile, instance_network_dropped_packets_count (RATE_LIMIT_DROP) confirms collector NIC PPS saturation, and the slight mirrored_bytes_count excess is recognized as normal Ethernet header inclusion.INGRESS or EGRESS, specific CIDRs, or specific protocols) cuts mirrored volume immediately.NONE (5-tuple hash) session affinity and backend subsetting disabled distributes flows evenly to resolve RATE_LIMIT_DROP saturation.Unlike disabling the policy, shrinking the collector fleet to one instance, or misconfiguring load balancer affinity and MTU, live filter tuning combined with compliant MIG horizontal scaling resolves both source-VM bandwidth pressure and collector NIC saturation with zero downtime.
LoadBalancer Service of collector Pods, enable Private Service Connect for the collector backend, and resolve backend health check failures as the cause of the packet drops.A Packet Mirroring policy links mirrored sources to a single regional internal passthrough Network Load Balancer forwarding rule, GKE LoadBalancer Services expose Kubernetes Pods, Private Service Connect (PSC) enables private service publishing across VPCs, and health checks monitor backend readiness.
LoadBalancer Service as a Packet Mirroring collector destination is not supported, and Packet Mirroring cannot mirror traffic for Private Service Connect published services./mirroring/dropped_packets_count or collector RATE_LIMIT_DROP errors.Separating traffic across two collector pools is needed using distinct, non-overlapping filter rules (for example, TCP to one Compute Engine MIG collector and UDP to another), rather than identical policies or GKE LoadBalancer Services.
When collector VMs experience RATE_LIMIT_DROP, the bottleneck is PPS or CPU capacity on the collector NICs rather than load balancer health check status.
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