Performance Test Novice Misunderstanding _ test

People who have some experience in performance testing are easy to get into this state, they are already familiar with the basic process of performance testing and are able to work with the test tools more skillfully. I was probably in the performance test about a year or so when I encountered this problem, I felt that the performance test process does not have too many challenges, every system encountered, as if the same process can be completed. Half a day to fill out the test plan, one time to prepare the test environment, a day to prepare test scripts, a few days to complete a variety of test cases (benchmark test, daily stress test, peak stress test, absolute concurrency test, stability test, etc.), then tuning, problem retest and completion of test reports. In my opinion, performance testing seems to have become a tool to perform a fixed use case.

After this kind of work lasted for a while, I felt something was wrong, there must be something wrong. Is the performance test really so simple as to put any one system into the standard process. So I began to think about the meaning of the test, and why it was tested for performance. It's because performance testing can provide information about bottlenecks, flaws, efficiencies, and so on that we think are valuable. That's just a tool, or a fixed process, to find out about the different systems. This is clearly not possible.

I began to try to understand the system as far as possible, the purpose of the system, the user is how to use the system, the user of which business performance is more sensitive, the system of some of the key business implementation logic is what, from the design implementation point of view which business performance may have hidden trouble. These are very few technical problems, need to do is to think, and then think deeply. Slowly I got some harvest, and began to understand why to test this system, what is the most important for this particular system, in order to get the information I need to test from which aspects, in order to realize my ideas and need which methods or tools. (now in my performance testing process, the time invested in understanding the system, understanding the user, and organizing the test ideas has increased significantly.) How much did you put in? ）

It's really hard to do these things, and every system that's been tested seems to me to be a new challenge. But gradually I found myself thinking more comprehensive, can quickly grasp the focus of the system, to find more important bugs, the actual performance of the system has been more accurate assessment. Here's a simple question of how to make sure your test results are consistent with the performance of your production environment, which means that the test results can actually reflect actual performance, not just the performance of the few test scenarios you've chosen. In other words, but please think about it carefully, how much you are sure of it.

It's just a few personal impressions, and I think if there's no way to get to a new level on the "mind", your performance-testing career may have reached a "bottleneck". How to break through this bottleneck, it requires efforts to change their own, more thinking and learning, the core of the ability is not capable of training. There must be some people think that the key to performance testing is "technology", so they constantly remember the parameters of various tuning configuration, that they mastered the essence of performance, as if what the system to their hands, as long as a few parameters will appear miracles. I have also experienced this stage, but also had several times to think very clever tuning experience, but also for their own will be a variety of middleware database configuration tuning and some small complacent. But now think about it, it is really a relatively low level, the ideological grasp of the key, not to see the overall situation, technically is just a little fur. In the face of such people, just ask a few why they can not answer, why to tune. Why do you want to tune this argument. How to prove the effect of this adjustment.

The simple summary above is a few things, hoping to give the performance test novice a little help:

1. The difficulty of performance testing lies in the understanding of the system, which is the analysis of the test pilot. In order to realize the idea of testing, there can be many ways, the means is always auxiliary, only thought is fundamental. Tools (such as LR) are not equal to performance testing, do not assume that the LR will understand the performance test, which is only the lowest level of test execution. Also do not think that will adjust a few parameters to understand the performance test, it is also a relatively low level.

2, tune the technology is not the main purpose of performance testing, good performance is not transferred out. Testers must understand the value of their existence, the so-called "technology" is only to achieve their own testing purposes of some means, compared with developers, DBAs, you are always a layman in these technologies.

3, do not follow the document template, fill in the test plan. Each system is different, to truly realize this, for each system to design a targeted test plan. Think about the meaning and purpose of each step of your work.

4, how to prove the effectiveness of test results is a difficult problem, it is worth taking time to seriously think. This process involves some very important content, such as the establishment of user models, followed by a dedicated article.

5, performance testing is a need for continuous improvement of the process, each time just try to do better, do a little bit of the past did not think of things. By accumulating, you will find yourself having a deeper understanding of the performance test.

Copyright Notice: This article is from the msnshow 51Testing Software test blog, see more: http://www.51testing.com/html/news.html