How to Achieve 90% Confidence in VPC Peering Packet Loss Metrics?

You have two VPCs: VPC A in Project A and VPC B in Project B. The VPCs are peered, and each VPC has VM instances in four zones. You are using the Network Intelligence Center Performance Dashboard to investigate the packet loss for traffic flows that start in VPC A and terminate in VPC B. You need the reported packet loss metric to have at least a 90% confidence level. What should you do?

Ensure that each zone in each of the VPC networks has at least 10 compute instances. Look in Project A for the reported metric.
Ensure that each zone in each of the VPC networks has at least 9 compute instances. Look in Project B for the reported metric.
Ensure that each zone in each of the VPC networks has at least 9 compute instances. Look in Project A for the reported metric.
Ensure that each zone in each of the VPC networks has at least 10 compute instances. Look in Project B for the reported metric. Source Reference Answer

Community Votes

71%

29%

71% of anonymous learners picked answer D. Votes are pick records left by other test-takers — they are not the verified answer.

Community Insight

Tests the NIC dashboard sampling formula and destination-project metric visibility, with the common trap being incorrect rounding or checking the source project instead of the destination.

This question evaluates the statistical sampling thresholds and project visibility rules for Google Cloud's Network Intelligence Center packet loss metrics. Community consensus confirms that deploying 10 VMs per zone and monitoring the destination project is mandatory to meet the 90% confidence requirement.

Option B is frequently chosen due to a calculation error where candidates assume 9 VMs suffice, failing to recognize that the official formula strictly requires 2.5 multiplied by the number of zones, resulting in exactly 10 VMs per zone.

Community Discussion (9 comments)

Positron75 👍 5 Selected: D

D, as explained in https://cloud.google.com/network-intelligence-center/docs/performance-dashboard/concepts/metrics-views?hl=en#metric_availability_and_confidence_levels and https://cloud.google.com/network-intelligence-center/docs/performance-dashboard/concepts/metrics-views?hl=en#packet-loss "90% confidence 2.5 VMs multiplied by the number of zones in the project." Which in our case is 2.5 x 4 = 10. "If the peered networks are in different projects, packet loss is visible in the destination project." Which in this case is project B.

anshad666 👍 1 Selected: D

D only

irmingard_examtopics 👍 4 Selected: D

For a 90% confidence, the required number of VMs in each zone is 2.5 VMs x the number of zones in the project, so 10 VMs per zone. Packet loss is visible in the destination project.

toasterbath777 👍 1 Selected: A

https://cloud.google.com/vpc/docs/monitoring-packet-mirroring

toasterbath777 👍 1

If it's 2.5 x 4 zones wouldn't that be 10? 2.25 x 4 zones equals 9 which is not what the link states. Are you factoring in the 90% to get to 9?

desertlotus1211 👍 1

90% confidence 2.5 VMs x the number of zones in the project. For example, if you have 12 zones in your project, you must have 30 VMs in each zone.

desertlotus1211 👍 1

Anything with VPC A as the destination is wrong: On the other hand, suppose that these two networks are not in the same project. That is, suppose network A is part of project A, and network M is part of project M. When the networks are peered, project M's Performance Dashboard shows packet loss data for situations where zone M is the destination zone. Conversely, when zone A is the destination zone, the packet loss data is visible only to project A

[Removed] 👍 4 Selected: B

To achieve a reported packet loss metric with at least a 90% confidence level, you need to ensure that the sample size is adequate for statistical significance. In this case, the question mentions that the packet loss investigation is for traffic flows starting in VPC A and terminating in VPC B. Since the traffic is originating from VPC A and terminating in VPC B, you should look in Project B for the reported metric, as this is where the traffic terminates. This makes options A and C incorrect. Now, you need to ensure that each zone in each of the VPC networks has at least 9 compute instances to provide a sufficient sample size for statistical analysis with a 90% confidence level. Therefore, the correct answer is: B. Ensure that each zone in each of the VPC networks has at least 9 compute instances. Look in Project B for the reported metric.

[Removed] 👍 2

"The metrics for the collector destination provides an aggregate view of all mirrored traffic that it collects. However, for the individual mirrored instances, you can view the number of mirrored packets that were dropped. Metrics for the collector destination don't include dropped packets." https://cloud.google.com/vpc/docs/monitoring-packet-mirroring

Comments & Corrections

No comments yet — spotted an error or have a note? Share it below.

Expert Analysis

Why the Answer Is Correct

Google Cloud’s Network Intelligence Center requires a minimum sample size to ensure statistical reliability for its performance metrics. For a 90% confidence level, the documentation specifies multiplying 2.5 by the total number of zones in the project, which yields 10 VMs per zone in this scenario. Additionally, packet loss metrics for VPC peering are collected and displayed in the destination project because the mirrored traffic terminates there, making Project B the correct location to view the data. As noted in community comments, this aligns precisely with the official sampling guidelines and collector behavior.

Why the Other Options Are Wrong

Options A and C incorrectly direct you to Project A, but the performance dashboard only aggregates packet loss at the destination endpoint where traffic is terminated. Option B fails the mathematical requirement by suggesting 9 VMs per zone, which falls short of the 2.5 × 4 = 10 threshold mandated for 90% confidence. Selecting either the wrong project or an insufficient sample size will result in incomplete or statistically unreliable metrics in the console.

Community Comment Notes

Multiple highly upvoted comments validate the 2.5 × zones formula and explicitly cite the official documentation for metric availability and confidence levels. Comment [8] provides a clear explanation of how cross-project peering routes packet loss visibility exclusively to the destination project. Comment [6] initially questions the rounding but ultimately confirms that 10 VMs is the correct threshold per Google’s guidelines. Collectively, these discussions reinforce that strict adherence to the sampling math and destination-project visibility is essential for passing this exam objective.

Official Reference

Exam Strategy

Memorize the Network Intelligence Center sampling thresholds (90% confidence requires 2.5 VMs per zone, while 95% requires 4 VMs per zone) and always determine traffic direction before selecting monitoring locations. When troubleshooting cross-project VPC peering, remember that performance metrics like packet loss are consistently aggregated at the destination project’s dashboard.

Related Analysis

Practice All PCNE Questions

Access 80 questions with complete answers and detailed explanations.

View Full PCNE Practice Test →

← Back to PCNE Study Guide