0

I have an application and system that exposes its resource metrics to Prometheus and Grafana that I want to do performance testing on. I want to be able to conduct a test, capture the metrics on CPU, Memory, File I/O, etc, for a test run, and save them for comparison against another test run. I'll be running multiple tests at different times with different configurations but the resource metrics will be standardized.

For Example, for two 5 minute tests, Comparing CPU:

  • Test 1 - 0:00-5:00
    • CPU Graph
  • Test 2 - 0:00-5:00
    • CPU Graph

And then Comparing Memory:

  • Test 1 - 0:00-5:00
    • Mem Usage Graph
  • Test 2 - 0:00-5:00
    • Mem Usage Graph

Is there a way to save or record these metrics for A/B comparison using Prometheus and Grafana? Or would I be forcing these tools too hard to do something that they really aren't designed to do? If so, is there a better way/tool to capture and compare the data I'm looking for this type of performance testing?

1 Answer 1

1

Did you find any usable solutions?

Would it be enough accurate to having one database for gathering data and then making comparing results by using test start and end time? Grafana has option for directly pointing certain timeframe. https://grafana.com/docs/grafana/latest/dashboards/time-range-controls/

I assume grafana and prometheus together are mostly usable for continues performance monitoring, data anylysing and alerting (and for multiple data sources). The solution I told would be one I will use to compare performance test results with same tools than you mentioned, if I don't find better.

Web server testing with general API's has multiple load testing tools and probably ability to compare performance metrics of certain test scenario. Those tools does not make what I want from testing purpose, so I am wondering same question than you have.

1

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.