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 troubleshooting intermittent application timeouts using Performance Dashboard in Network Intelligence Center. Over 24 hours, you observe four patterns across Per Project and Google Cloud views:
| Path | Metric | Observed Visualization |
|---|---|---|
us-central1-a to us-central1-b | Packet loss | Heatmap shows 3.1%* for 3 minutes during preemptible VM shutdowns; Google Cloud is 0%. |
us-central1-c to europe-west1-b | Packet loss | Time-series shows 6.4% for 8 minutes; Google Cloud average is 5.9%. |
europe-west1-b to asia-east1-a | Latency | Time-series displays a flat line at 255 ms for 3 hours. |
us-central1-a to Internet | Latency | Low-traffic TCP service spikes to 420 ms; ping shows 32 ms. |
Which observation indicates critical underlying Google Cloud network degradation rather than normal variation, low sample confidence, or an application-level artifact?
In Performance Dashboard, sustained packet loss on a bidirectional zone-pair time-series chart that exceeds predefined severity thresholds and aligns with the Google Cloud performance view indicates genuine underlying network degradation rather than local project noise or transient probe variance.
This pattern satisfies every diagnostic criterion for an infrastructure-level network incident:
6.4% packet loss rate exceeds the 5% for 5 minutes threshold used by Google Cloud's predefined High packet loss alert policy, signaling severe impact to TCP throughput and application timeouts.us-central1-c to europe-west1-b.6.4%) with the Google Cloud average (5.9%) at the exact same timestamp and zone pair confirms that the issue affects Google Cloud infrastructure broadly rather than being isolated to a project-specific configuration or low-confidence sampling error.monitoring.timeSeries.list.Unlike the other observations—which reflect low-confidence preemptible VM noise (3.1%*), normal latency distribution bucketing (255 ms), or TCP passive-sampling artifacts (420 ms vs. 32 ms ping)—only this observation demonstrates sustained, high-magnitude, globally corroborated packet loss.
Performance Dashboard derives latency (RTT) metrics through passive sampling of actual TCP traffic, calculating the interval between a sent TCP sequence number (SEQ) and the corresponding acknowledgment (ACK). Because this measurement occurs over Layer 4 packets, both network transit time and the remote endpoint's TCP/IP stack or application response behavior can influence the sampled values.
This spike is a measurement and application-level artifact rather than an underlying Google Cloud network issue:
1,000 TCP packets per minute threshold), Performance Dashboard relies on 60-second probes. When a TCP service responds quickly with a bare ACK followed later by an application payload, packet sampling can miss the fast ACK and mistakenly pair the earlier SEND with the delayed application response as the closing ACK.ping test measuring 32 ms RTT bypasses TCP application-layer processing and proves the Layer 3 network path between us-central1-a and the internet client is performing normally.1,000 TCP packets per minute for reliable visibility, whereas Google Cloud global latency aggregates across all projects.ping or application tracing before escalating a network incident.Both the Performance Dashboard TCP latency graph and the independent ICMP ping test showed a matching 420 ms RTT increase, and the Google Cloud view confirmed elevated latency for us-central1 to that geographic region.
In Performance Dashboard, latency (round-trip time, or RTT) is measured passively from sampled TCP traffic headers and stored as a percentile distribution grouped into fixed histogram buckets (such as 10 ms intervals: 0–10 ms, 10–20 ms, up to 250–260 ms). The dashboard plots the median (50th percentile) RTT derived from those distribution buckets.
A completely flat, unchanging horizontal line at 255 ms over several hours is a normal visualization artifact of distribution bucketing, not a frozen metric or sudden network outage:
251 ms, 253 ms, and 259 ms all falling inside the 250 ms–260 ms bucket), the chart displays the arithmetic mean of that bucket interval (255 ms).europe-west1-b (Belgium) and asia-east1-a (Taiwan) naturally exhibits baseline propagation latency in the 250 ms range due to physical fiber distance.±4 ms) remains wholly contained within a single 10 ms bucket.SEQ packet and its corresponding ACK.The latency time-series between europe-west1-b and asia-east1-a showed an abrupt, sustained jump far above its historical 6-week baseline (for example, rising from 255 ms to 400+ ms) and correlated with a similar elevation in the Google Cloud view.
In Performance Dashboard, packet loss is measured using active probing between virtual machine (VM) physical hosts across the project's VPC mesh, and zone-pair heatmaps annotate each cell with statistical confidence indicators. An asterisk (*) next to a percentage indicates low measurement confidence resulting from an insufficient number of active probes per minute—typically caused by having too few VMs deployed relative to the number of zones in the project.
This observation reflects transient noise and expected operational behavior rather than a critical Google Cloud infrastructure failure:
0% packet loss for us-central1-a to us-central1-b, confirming the underlying regional fabric is healthy.10 × (number of zones) VMs per zone, while 90% confidence requires 2.5 × (number of zones) VMs per zone.*), and zone pairs lacking enough probes to produce meaningful data display N/A.The packet loss between us-central1-a and us-central1-b exceeded 5% for more than 5 consecutive minutes without an asterisk (*) confidence warning, occurred independently of preemptible VM shutdowns, and aligned with elevated packet loss in the Google Cloud performance view.
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