2

I'm trying to get a baseline benchmark of an API before making some changes and I've been given the requirement that the changes can introduce a maximum of 1 millisecond of additional time to each request. Taurus/JMeter seems to only be able/willing to show me down to the millisecond, so I'm concerned that rounding will now become a factor in determining success/failure. Sure, averaging multiple runs is better than basing decisions on a single run, but I'd also rather things be at least somewhat deterministic (i.e. no "luck of the draw" affecting success or failure).

Is anyone familiar with getting more granularity out of Taurus/JMeter? Is this a situation where I may need to ditch Taurus and use JMeter directly? Or do I need to abandon both and go with another tool? Microseconds or nanoseconds would be fantastic, but even just a single extra decimal point (0.0143s vs 0.014s) would be an improvement.

David Young
  • 441
  • 1
  • 4
  • 13

1 Answers1

1

JMeter goes as of 4.0 up to millis not more in the HtML report which is not the same as the one generated by Taurus.

Did you try without Taurus to see what you get in html report:

In short, You just need to run:

jmeter -n -t test.jmx -l results.csv -e -o report_folder

If it’s not more precise than Taurus, you’ll need to go for another tool.

UBIK LOAD PACK
  • 33,980
  • 5
  • 71
  • 116