Metrics Fundamentals

Overview

Metrics provide numerical signals about system behavior over time, enabling trend analysis and alerting.

Metric Types

• Counter: cumulative increasing value
• Gauge: current measured value
• Histogram: distribution of observed values
• Summary: quantile-oriented distribution

flowchart LR
    A[Application] --> B[Metric Endpoint]
    B --> C[Collector]
    C --> D[Time Series Store]

Practical Example

http_requests_total{service="orders",status="200"} 84521
request_duration_seconds_bucket{le="0.5"} 10234

Production Example

Teams monitor latency percentiles and error rates to detect degraded customer-facing behavior before outages.

Common Pitfalls

• High-cardinality labels causing storage pressure
• Tracking too many metrics without actionability
• Missing clear ownership for service-level indicators

Key Takeaways

• Metrics are foundational for proactive monitoring.
• Label design strongly affects metric scalability.
• Focus on signals tied to user and business impact.