Metrics and dashboards in GCP - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ service_behavior

intermediate

Understanding Metric Types in Google Cloud Monitoring

Which metric type in Google Cloud Monitoring is best suited to track the total number of requests received by a web server over time?

A Cumulative metric, because it counts the total number of events since a start time.

B Gauge metric, because it measures the current value at a point in time.

C Delta metric, because it measures the change between two points in time.

D Distribution metric, because it shows the distribution of latency values.

💡 Hint

Think about a metric that keeps adding up over time without resetting.

A cumulative metric counts the total number of events from a start time, making it ideal for tracking total requests over time. Gauges show current values, deltas show changes between points, and distributions show value spreads.

❓ Architecture

intermediate

Designing a Dashboard for Latency Monitoring

You want to create a Google Cloud Monitoring dashboard that shows the 95th percentile latency of an API over the last hour. Which chart type and aggregation method should you use?

A Use a line chart with 95th percentile aggregation over the last hour.

B Use a bar chart with sum aggregation over the last hour.

C Use a line chart with mean aggregation over the last hour.

D Use a pie chart with max aggregation over the last hour.

💡 Hint

Percentiles help understand latency distribution, and line charts show trends over time.

A line chart with 95th percentile aggregation shows how latency behaves at the high end over time, which is useful for performance monitoring. Mean or sum do not capture percentile behavior, and pie charts are not suitable for time series.

❓ security

advanced

Securing Access to Monitoring Dashboards

Which IAM role should you assign to a user to allow them to view Google Cloud Monitoring dashboards but prevent them from modifying any metrics or alert policies?

A roles/owner

B roles/monitoring.viewer

C roles/monitoring.admin

D roles/monitoring.editor

💡 Hint

Look for the role that grants read-only access to monitoring data.

The roles/monitoring.viewer role allows users to view dashboards and metrics without permission to change configurations. Editor and admin roles allow modifications, and owner has full project control.

❓ Configuration

advanced

Configuring Alerting Policies for CPU Usage

You want to create an alerting policy that triggers if the average CPU usage of a Compute Engine instance exceeds 80% for 5 continuous minutes. Which configuration correctly sets this condition?

A Metric: compute.googleapis.com/instance/cpu/utilization, Aggregation: max, Condition: above 80 for 5 minutes

B Metric: compute.googleapis.com/instance/cpu/usage_time, Aggregation: sum, Condition: above 80 for 5 minutes

C Metric: compute.googleapis.com/instance/cpu/utilization, Aggregation: mean, Condition: above 0.8 for 5 minutes

D Metric: compute.googleapis.com/instance/cpu/usage_time, Aggregation: mean, Condition: above 0.8 for 5 minutes

💡 Hint

CPU utilization is a ratio between 0 and 1, so thresholds should match that scale.

The metric compute.googleapis.com/instance/cpu/utilization measures CPU usage as a fraction (0 to 1). Using mean aggregation and threshold 0.8 (80%) for 5 minutes matches the requirement. Usage_time is cumulative CPU time, not a percentage.

✅ Best Practice

expert

Optimizing Dashboard Performance with Large Metrics Data

You manage a dashboard with multiple charts showing high-cardinality metrics from thousands of instances. The dashboard loads slowly. Which approach best improves dashboard performance without losing critical data?

A Increase the refresh rate to update charts every 10 seconds instead of every minute.

B Reduce the time range displayed to the last 1 hour instead of 7 days.

C Remove percentile aggregations and use only mean aggregations for all charts.

D Use metric filters to aggregate data by instance group instead of individual instances.

💡 Hint

Aggregating data reduces the number of time series and improves load times.

Aggregating metrics by instance group reduces the number of time series, improving dashboard load times while preserving meaningful data. Reducing time range helps but may lose historical context. Removing percentiles loses important insights. Increasing refresh rate worsens performance.