Performance Testing Mistakes Techies Must Avoid

If you are a software engineer actively involved in the performance testing segment of software product lifecycle, here is what you must avoid doing.
Mistake 1: Confusing bulk loading with performance testing
It has been observed that software engineers are often confused over performance testing and load testing. The fundamental approaches for both of them are different.
While load testing is bulk loading of the infrastructure to understand how much volume the system can process, performance testing is how fast the system is working while under heavy load.
For effective performance testing, the ideal route is to put the system through a series of simulated activities. Simulate various browsers, networks and user types, because systems are adept at quick response but not so good at level setting.
Mistake 2: Treating all users the same
Understand that different users place different types of load on the system since the expectation levels vary across different levels. For example, a patient user may give about five seconds to a website before it loads while an impatient user may not even give one second. So developers will have to understand that such user behaviour will need to be reflected in the difference in approaches in terms of the backend integration points for different users.
How to address this:
  • Remember that key is prioritising performance based on user types and devices
  • So first segment user types and then use this to prioritise services and isolate problems better
Mistake3: Not associating snapshots to releases
With increasingly complex web apps, even the slightest change in the app can impact the performance. For example, even the change in placement of the company logo on the webpage can possibly impact the performance of the website.
So, one needs to keep a close tab on even the slightest of changes to make it easier to roll back to root causes. This leaves room for improvement and writing efficient codes.
This can be addressed by adding performance testing to continuous integration processes show performance trends related to updates to the app. Feedback loop can isolate bad/slow code.