Skip to main content
PromGleam provides support for the core Prometheus metric types. Understanding which metric type to use is essential for effective monitoring and observability.

​
Metric types

Prometheus defines four core metric types that serve different purposes:
1

Counter

A cumulative metric that represents a single monotonically increasing counter whose value can only increase or be reset to zero on restart. Use counters for metrics like:
  • Total number of HTTP requests served
  • Total number of tasks completed
  • Total number of errors
2

Gauge

A metric that represents a single numerical value that can arbitrarily go up and down. Use gauges for metrics like:
  • Current memory usage
  • Number of concurrent requests
  • Temperature readings
  • Cache size
3

Histogram

A metric that samples observations (usually things like request durations or response sizes) and counts them in configurable buckets. It also provides a sum of all observed values. Use histograms for:
  • Request duration
  • Response sizes
  • Query execution times
4

Summary

Similar to histograms but calculates configurable quantiles over a sliding time window. Summaries are not yet supported in PromGleam.
Do not use a counter to expose a value that can decrease. For example, do not use a counter for the number of currently running processes; instead use a gauge.

​
Registries

All metrics in PromGleam are registered in a named registry. The most common registry name is "default", but you can create and use multiple registries to organize your metrics. 
import promgleam/metrics/counter.{create_counter, increment_counter}

// Create a metric in the default registry
create_counter(
  registry: "default",
  name: "http_requests_total",
  help: "Total number of HTTP requests",
  labels: ["method", "route", "status"],
)

​
Labels

Labels allow you to partition your metrics into multiple dimensions. When you create a metric, you specify the label names. When you record a measurement, you provide the label values.
The order of label values must match the order of label names defined when creating the metric.
import promgleam/metrics/counter.{create_counter, increment_counter}

// Define labels when creating the metric
create_counter(
  registry: "default",
  name: "http_requests_total",
  help: "Total number of HTTP requests",
  labels: ["method", "route", "status"],
)

// Provide label values in the same order when recording
increment_counter(
  registry: "default",
  name: "http_requests_total",
  labels: ["GET", "/api/users", "200"],
  value: 1,
)

​
Naming conventions

Follow these best practices when naming your metrics:
  • Use snake_case for metric names
  • Include the unit of measurement in the metric name (e.g., _seconds, _bytes, _total)
  • Use descriptive names that clearly indicate what is being measured
  • Add _total suffix to counters
For time-based metrics, use seconds as the base unit. Prometheus best practices recommend using base units (seconds, bytes, etc.) rather than multiples (milliseconds, kilobytes, etc.).

​
Next steps

Counters

Learn how to use counter metrics

Gauges

Learn how to use gauge metrics

Histograms

Learn how to use histogram metrics

Registries

Learn how to manage registries

Build docs developers (and LLMs) love

Get started for freeTalk to us