Which key observability signal helps detect real-time performance bottlenecks in a Kubernetes cluster?

Metrics are numerical, time-series data representing the state of a system, such as CPU utilization, memory consumption, or request latency. They are lightweight, easily aggregated, and ideal for real-time visualization on dashboards and for triggering alerts. This continuous, quantitative data stream allows operators to immediately observe trends, spikes, and saturation points that indicate performance bottlenecks across the Kubernetes cluster. Tools like Prometheus are designed specifically to collect and analyze these metrics for real-time performance monitoring.

A. Logs are verbose, discrete event records used for deep-dive debugging of specific incidents, not for an aggregated, real-time overview of system performance.
B. Traces follow a single request's path through distributed services, pinpointing application-level latency, but not providing a view of overall cluster infrastructure health or resource bottlenecks.
D. Events are Kubernetes objects that report on discrete lifecycle occurrences (e.g., a pod being scheduled). They are reactive symptoms, not the continuous data needed for proactive bottleneck detection.

1. Kubernetes Official Documentation, "Monitoring, Logging, and Debugging": The documentation distinguishes between different observability signals. In the section on "Monitoring," it states, "Often, metrics are numeric time series data... In a cluster, metrics are useful for understanding system state." This highlights metrics as the primary tool for understanding real-time system performance. (See: kubernetes.io/docs/tasks/debug/debug-cluster/logging-monitoring-troubleshooting-overview/)

2. OpenTelemetry Documentation, "Metrics": As a core CNCF project for observability, OpenTelemetry defines metrics as "a measurement about a service, captured at runtime." It explains that metrics are aggregated to provide statistical insights into system behavior, which is fundamental for identifying performance issues. (See: opentelemetry.io/docs/concepts/signals/metrics/)

3. Armbrust, M., et al. (2010). "A view of cloud computing." Communications of the ACM, 53(4), 50-58. This foundational academic paper on cloud computing discusses the importance of monitoring for elasticity and performance. It emphasizes the need to "monitor resource usage in real time to decide when to request and relinquish resources," a task accomplished through metrics. (DOI: https://doi.org/10.1145/1721654.1721672, Section 4.1, "Measuring and Monitoring Utilization").