The Prometheus documentation about histogram_quantile uses the following query as an example to get the 90th percentile of the duration of the requests over 10m.
histogram_quantile(0.9, rate(http_request_duration_seconds_bucket[10m]))
But why would I want to calculate this over a timespan? If I omitted the rate and the range vector like below, wouldn't I get a more accurate measurement?
histogram_quantile(0.9, http_request_duration_seconds_bucket)
I have read other answers about the topic, but I still can't understand what does the rate function do in a histogram_quantile.
question from:https://stackoverflow.com/questions/66050834/what-is-the-effect-of-using-rate-in-histogram-quantile
