Prometheus supports two kinds of complex metrics: Summaries and Histograms.
These metrics are being used to keep track of the number of observations and the sum of observed values. It generates time series in the database. For example, they all add the suffix _sum to the observed value’s sum.
Histogram
A histogram is used to represent the counts and observations (typically response size and request durations) in the configuration buckets. It also makes the sum of each observed value easier.
It makes a histogram an important choice for tracking things like latency, which may have SLO (Service Level Objective) defined across it.
Summary
A summary is used to represent different observations (like response size or request durations usually). It also displays the total number of observations and the sum of each observed value. On any sliding time window, it can calculate configurable quantities.
