90% Percentile in Performance Testing

                               90% Percentile in Performance Testing

Let’s talk about performance testing concept in general. One of the most important concepts which we always come across after completing load test is 90% Percentile Response time.

What is Percentile?

Let’s say 60% of people are shorter than you: percentile 60%. That means you are at the 60th percentile.

i.e. Percentile relates to where an individual item stands within a population/a bulk of data set.

Now we come to 90%Percentile

Let’s consider an example:

We execute a load test of 20 iterations. We captured the response times of 20 iterations as below:

No.	Iteration No.	Login (Response Time)
1	1	0.8
2	2	1.5
3	3	1.8
4	4	1.1
5	5	1.6
6	6	1.7
7	7	1.3
8	8	0.9
9	9	1
10	10	1.5
11	11	2.3
12	12	1.9
13	13	1.5
14	14	1.6
15	15	1.1
16	16	0.9
17	17	2.1
18	18	1.9
19	19	1.4
20	20	1.9

Now sort this data set from small to higher range:

No.	Iteration No.	Login (Response Time)
1	1	0.8
2	2	0.9
3	3	0.9
4	4	1
5	5	1
6	6	1.1
7	7	1.1
8	8	1.3
9	9	1.4
10	10	1.5
11	11	1.5
12	12	1.5
13	13	1.6
14	14	1.6
15	15	1.7
16	16	1.8
17	17	1.9
18	18	1.9
19	19	2.1
20	20	2.3

Now 90% of my total transactions is 20*(90/100) = 18 so the 18th value will be 90th percentile which is 1.9 seconds.

This is how we calculate the 90 %percentile.

1. Sort the traction response time by smaller to large values

2. Remove the top 10 % instances

3. The highest vale left is the 90^th Percentile.

Why 90 % Percentile is important when we can get average response time.

Averages are not effective enough since they are too simplistic and one-dimensional.

whereas Percentiles are easy way of understanding the real performance characteristics of application. Percentiles also provide a great basis for automatic baselining, behavioral learning and optimizing application with a proper focus. In short, percentiles are major factor which gives clear picture which transactions require tuning.

Happy Testing!!