Prometheus
Counters
- When you want to record a value that only goes up
- When you want to be able to later query how fast the value is inreasing
- Usecase: request count, tasks completed, error count
Gauge
- When you want to record a value that can go up and go down
- When you don't need to query its rate
- Usecase: Memory, queue size, number of request in progress
Histogram
- When you want to take many measurement of a value to later calculate average or percentile
- when you're not bothered about the exact value, but are happy with the approximation
- When you know what the range of values will be up front so can use the default bucket definitions or define your own
- Usecase: Request duration, response size
Summary: DIfferent between summary and histogram
- Summary quantiles are calculated on the application server. Histogram quantiles are calculated on prometheus server
- Histogram request upfront bucket definition. Summaries are a good option if you don't know what the range of values will be upfront.