0

my average response time is coming greater than the 99 percentile( in JMeter).

James Z
  • 12,209
  • 10
  • 24
  • 44
Sneha Kamath
  • 11
  • 2
  • At least I don't understand what you're asking here -- and you figure even changes from 95 to 99. We don't see your screen or know what you're working with, so you should probably add a lot more info – James Z Nov 28 '18 at 14:10
  • A related question about averages and percentiles: https://stackoverflow.com/questions/40874994/loadrunner-analysis-how-can-the-90th-percentile-be-higher-than-the-average/55184413 – AdrianHHH Mar 29 '19 at 15:50

2 Answers2

1

So what? My too.

enter image description here

This is normal situation when you have:

  • low number of samplers with high response time

and

  • high number of samplers with low response time

In above setup I used 1 Dummy Sampler with 1000 milliseconds response time and 100 Dummy Samplers with 1 millisecond response time resulting in average response time of 11 ms and 99% of 4 ms.

More information:

Dmitri T
  • 159,985
  • 5
  • 83
  • 133
0

To get such results your response times above the 95th and 99th response time should be really high. For example, Let's say you have 100 requests First 94 samples have a response time of 20ms From 95th to 99th Sample the response time is: 30ms And the 100th request has a response time of 1500ms

If you take the average response time It'll be: ((94 * 20) + (30 * 5) + (1000 * 1))/ 100 Which is: 35.5ms But the 90th and the 99th percentile values will be 30ms.

It's not a normal situation as per my experience. It's better if you can investigate what caused the system to respond in unexpected low response time and fix the issue. (Try to run the script again and see if you can recreate the scenario too)

Note: The response times worse than the 99th percentile matters a lot

For more information: https://softwaretestresearch.blogspot.com/2019/03/qa-engineering-analysis-what-happened.html

Muditha Perera
  • 1,216
  • 8
  • 24