Which pillar would be most useful for predicting future resource usage?

Which pillar would be most useful for predicting future resource usage?

• A. Logs
• B. Traces
• C. Metrics
• D. Alerts

Answer: (C)

Predicting future resource usage hinges on time-series data that shows how usage changes over time. Metrics provide exactly that: numerical, time-stamped measurements like CPU utilization, memory usage, disk I/O, and network throughput collected at regular intervals. This data reveals trends, seasonality, and anomalies, enabling forecasting and informed capacity planning or autoscaling decisions. Logs are event records useful for troubleshooting but aren’t organized for trend analysis. Traces show the path and timing of individual requests, which helps with latency and debugging rather than overall resource forecasting. Alerts are reactive signals that trigger when a threshold is crossed; they indicate issues, not predictive capacity data.

Predicting future resource usage hinges on time-series data that shows how usage changes over time. Metrics provide exactly that: numerical, time-stamped measurements like CPU utilization, memory usage, disk I/O, and network throughput collected at regular intervals. This data reveals trends, seasonality, and anomalies, enabling forecasting and informed capacity planning or autoscaling decisions. Logs are event records useful for troubleshooting but aren’t organized for trend analysis. Traces show the path and timing of individual requests, which helps with latency and debugging rather than overall resource forecasting. Alerts are reactive signals that trigger when a threshold is crossed; they indicate issues, not predictive capacity data.