- Instrumenting your code
- Libraries
- Official: Go, Java/Scala, Python, Ruby
- Unofficial: Bash, C++, Common Lisp, Elixir, Erlang, Haskell, Lua for Nginx, Lua for Tarantool, .NET/C#, Node.js, PHP, Rust
- No client library available?
- Implement it yourself in one of the supported exposition formats
- Exposition formats:
- Simple text-based format
- Protocol-buffer format (Prometheus 2.0 removed support for the protocol-buffer format)
 Image Added
- 4 types of metrics
- Counter
- A value that only goes up (e.g. Visits to a website)
- Gauge
- Single numeric value that can go up and down (e.g. CPU load, temperature)
- Histogram
- Samples observations (e.g. request durations or response sizes) and these observations get counted into buckets. Includes (_count and _sum)
- Main purpose is calculating quantiles
- Summary
- Similar to a histogram, a summary samples observations (e.g. request durations or response sizes).
- A Summary also provides a total count of observations and a sum of all observed values, it calculates configurable quantiles over a sliding time window.
- Example:
- Your need 2 counters for calculating the latency
- 1) total request(_count)
- 2) the total latency of those requests (_sum)
- Take the rate() and divide = average latency
|
