Performance Testing Basics

Performance Testing Basics.

Performance testing activity has long been considered a niche area by most software companies. This has been primarily due to the high cost of the tools involved. With the Open Source carving a space for themselves in the world of software tools, it is time companies realized that they too can afford and contribute to performance testing. They too are in a position to offer their services to their clients.

What is needed?
1) First, loads and loads of patience to go about the performance testing task.
2) The ability to understand the application and try and find out ways to conduct an effective test.
3) A bit of technical knowledge does help since it helps you to pinpoint and analyze the possible bottlenecks.



How it is done?

1) Initially the application is tested with 1 user. This is done to know the basic empty cache performance criteria. It is similar to starting
a car on a cold winter morning.

2) Once the cache is taken care of by having the single user test, the users are incremented in the following mode (10, 25, 50, 75, 100, 125 … in the increments of 25 or 50 as may be deemed fit).

NOTE : For each user, say for example 10 users, it is required to generate atleast 10 to 15 test outputs so that we know the application behaves consistently across the 10 or 15 readings. This has to be done for all the other set of users.

3) From these output readings, we arrive at the average and then record it separately.

4) Once the average response times are taken for all the set of users, we can plot a separate graph which shows the actual deterioration of the application at a particular point in time.

By the end of the testing cycle, we realize we have executed the script nearly 150 times or more in order to test the performance. But it is worth the effort to give us the required insight.

This is what the final report will resemble.


Most preferred Open Source Performance testing tools :
- OpenSTA
- JMeter