Performance test process (four)-test tool selection

Based on the analysis of the system performance test points, we choose the right job. Not all performance requirements have to be performed with LoadRunner or Jmeter. As long as the demand can be met, I personally prefer lightweight performance tools or write code by myself.

Comparison of test tools:

Software performance testing tools are divided into three categories, commercial charging tools, open source free tools, and independent development tools.

// Business tools Open source tools Self-developed tools
advantage Powerful and easy to use; provides rich functions, supports more protocols and concurrent users, rich counters and test results; automatically generates scripts, low learning costs; strong stability and reliability. Lightweight, flexible and easy to expand; free to download and use, you can view the source code of the tool, and easy to expand upon root needs. Close to the needs, develop tools that are more suitable for the needs, and have a lower learning cost than open source tools. A unique tool system can be formed.
Disadvantage It is difficult to expand depending on the functions provided by the tool itself. The tool is bulky and the purchase cost is high. The cost of learning is high, it can't fully meet the performance test requirements, and it needs function expansion. The function is relatively simple. Tool development costs are high, requiring professional tool developers or teams, and tool costs are also uncertain.

Through the above comparative analysis, it is difficult to judge the choice of a certain type of tool. Each has its own obvious advantages, and of course, there are also obvious shortcomings. How to choose a tool better depends on the actual performance requirements.

The following aspects need to be considered:

  1. Performance requirements:

The first thing to consider is whether the performance requirements are met. For example, the system protocol does not support, scripts cannot be created, the number of simulated users of the system cannot be reached, or certain special functions cannot be tested, such as streaming media performance testing. Some performance tools cannot meet these requirements, so of course they must be eliminated first.

  1. Purchase cost:

The purchase cost of commercial tools is also an issue that we need to consider. In fact, this issue is also an issue that the boss needs to consider, and the overall budget for performance testing costs. Of course, under the premise of meeting performance requirements, of course everything is simple. The service cost of genuine LoadRunner is very high. What about open source software? There are actually many open source performance testing tools, because they are all suitable for different demand scenarios, so their size is much smaller than commercial software, and of course their functions are relatively simple. But in some cases, they can fully meet certain performance requirements, or they can be met through the cooperation of several open source tools, or they can be met by extending open source tools. Extending open source tools puts higher demands on testers. Testers are required to have considerable research on the principle and source code of the tool. This is very helpful for the improvement of testers' automatic capabilities. Or your performance test project is special, no one is satisfied with all the performance tools on the market, and the commercial customization cost is too high. Or you want to build your own tool system. Considering the particularity and long-term needs of the company, a performance test tool can be developed independently for use. One-time development, permanent benefits (of course, maintenance, upgrades, expansions, etc.), and closer to your own needs. At present, many well-known large companies have their own performance testing tools for internal use.

  1. Personnel learning cost:

The last thing to mention is the cost of learning tools for personnel. In fact, the first consideration for testers when choosing performance testing tools is to start with the tools they will use, so that they can learn and live with the lowest control. If a tool is used, The testers learned the test scripts for half a year, and the boss had let you go. This is also where the requirements for performance testers are higher. Therefore, for performance testers, you'd better be familiar with a process commercial performance tool, an open source free performance tool, and also need to be familiar with common performance script development languages. This is the basic requirement.

Guess you like

Origin blog.csdn.net/Python_BT/article/details/108751047