Operations → Observability
Metrics
Numeric measurements collected over time to observe system behavior and health.
Motivation
Metrics are numeric measurements collected over time. They solve the problem of seeing trends, rates, saturation, and health at a glance.
Mental model
A metric is a time-series answer to questions like how many, how fast, how full, or how often.
Where it fits
Metrics belongs to the Operations track, inside Observability. It connects to nearby concepts through the graph relationships on this page and in the graph explorer.
Common mistakes
- Using averages that hide tail latency.
- Creating high-cardinality labels accidentally.
- Tracking vanity numbers instead of operational signals.
Related concepts
Use the metadata panel or return to the graph to explore concepts that this one requires or relates to.
Counter
A cumulative metric that increases as events occur and is commonly converted into a rate.
Gauge
A metric representing a value that can rise or fall, such as queue depth or memory usage.
Histogram
A distribution metric that groups observations into buckets and supports percentile estimation.