Metrics
A practical DevOps definition, with an example you can use.
Numerical measurements of system behaviour recorded over time.
An example you can picture
A service records request count, error count, and response duration. A chart reveals that error rate increased after a deployment.
The distinction that matters
An average can hide slow requests. Consider distributions and percentiles, and keep labels controlled so the number of distinct series does not grow without bounds.
Put it into practice
Choose a request counter and a duration measurement for a lab application. Sketch how you would detect an increase in errors without relying only on the total number of requests.
Connect the ideas: slo and observability. For the bigger picture, read the related learning guide.
Reference
Read the primary documentation for the underlying concept and further detail.