Why do you want to do performance testing?

question 1: why should we conduct performance testing? There are three reasons:

Sichuan. The level of developers is different. Some things written are of high performance and some are of low performance, so it needs to be tested uniformly.

2. The programming tools themselves have performance problems, and the software developed with such tools should also confirm whether it meets the performance indicators required by the requirements, such as how many seconds the response time should be controlled.

3. Performance testing and strength testing are all aimed at testing the stability of the system. If the stability is good, the quality of the software will be good, and more money will be spent.

question 2: why stress testing? It is normal to face pressure in three types of pressure jobs, but some jobs often face pressure, and the pressure is relatively high. For this reason, the interviewer designs corresponding stress interview questions for this kind of job application to test the candidate's ability to withstand pressure. These positions can be divided into three categories: the first category is middle and senior management positions, and he has to face communication pressure from up and down, left and right, inside and outside, and pressure from all sides at any time and place. The second category is the sales staff, especially the sales of big customers, who need to communicate with customers directly, and the needs of customers are changing. The third category is special professional and technical posts, and the environment they face changes instantly, which will produce pressure. Candidates for these three positions are likely to meet stress interview. Three types of pressure working environment In the real working environment, we will be under pressure for many times, so there are three types of situations from the pressure environment. The first category, emergency environment: it allows you to do something quickly, which is beyond the average and particularly urgent. The second category, contradictory environment: how to do things well in this contradiction? In particular, coordination work is faced with several or even more than a dozen kinds of conflicts in work elements. How to solve them? The third category, unfamiliar environment: psychologically speaking, everyone has their own comfort zone, and once they leave the comfort zone and enter the unfamiliar environment, they will have pressure, such as: new jobs, new leaders, new assignments that they have never been exposed to, and so on.

question 3: how to do a good job in performance testing? Hello, first of all, I appreciate your attitude. When I was recruiting new people at TestBird, many children felt that they knew how to use tools and how skilled the steps were, and they felt very good about themselves.

actually, I want to say that the focus of performance testing is not on learning performance testing tools.

Of course, you can also analyze the pressure points of the system, record the script with LR, set up users, do stress, analyze the results, and sort out the test report. Completed the whole process of performance testing. Then I say this performance test report is effective, but it is not necessarily useful.

why? Because in the performance test report, in your environment, you have measured such an effect. It's not adulterated. It's all true.

Why is it not necessarily useful? Do you know the system architecture? Do you know the operation mode and processing mechanism of database, middleware and front-end program? Do you know the network protocol? Know anything about the operating system? Are you familiar with the language of the development system, such as the internal mechanism of java JVM? These are all part of the system operation, which are affecting the performance of the system. If you don't know this, how can you make valuable and meaningful performance tests?

so, it's good to learn these performance testing tools, but this is only the first step. The performance results are just some data. Knowing what you are doing and why you are doing it, and giving something valuable after you finish it, is what you should practice slowly.

question 4: what should I master to do a good job in performance testing? This kind of question has been asked many times before, so this time I just sort it out and send a topic for my colleagues' reference. If you need to add, you are also welcome to leave a message to discuss. If you really want to do a good job in performance testing, you still need to learn a lot. Make a brief list. 1. Proficient in the basic concepts, processes and methodologies of performance testing, and understand performance engineering;

3. Solid basic knowledge of computer specialty, including computer composition principle, operating system, database principle and computer network principle;

4. Be familiar with at least one commonly used database product, such as SQL Server or Oracle, be able to perform general database management operations, be familiar with the use of SQL scripts, and be familiar with commonly used data tuning tools and commonly used counter；;

5. Be familiar with the principle of at least one operating system, either Windows or Linux, the architecture of the operating system, important basic concepts of the operating system, memory management, storage/file system, driver/hardware management, implementation and composition of network protocols, performance monitoring methods and principles, and familiar with commonly used counter；;

6. be familiar with at least one web server product, such as apache, and understand the general configuration and common counter；;

7. Be familiar with at least one application server product, such as tomcat, general configuration, common server performance monitoring methods and principles, and common counter；;

8. Be familiar with at least TCP/IP protocol and HTTP protocol, and at least have seen and understood the use and configuration of three-layer and four-layer switching or routers. Understand the commonly used counter； related to network performance;

9. Understand the deployment architecture and application architecture of general large enterprise applications;

1. Understand the architecture and optimization process of well-known large-scale web applications, super-large-scale websites with high concurrency, high traffic and high real-time response requirements;

11. be familiar with the basic knowledge of statistics, common analysis methods and experimental design methods, and understand the knowledge related to mathematical modeling;

12. be familiar with the business knowledge and user scenarios of exclusive industries, such as the business knowledge and user scenarios involved in the OSS system in the telecommunications industry and the business knowledge and user scenarios involved in the securities trading system;

13. Lots of practical performance testing and optimization experience;

14. Actively participate in discussions, exchanges and sharing in various circles and societies. I have thought of so much for the time being, so interested friends can discuss it together. I believe everyone has their own different experiences and feelings, and can share them with others for reference.

In addition, I have published a lot of performance testing materials before, ranging from entry-level articles to required reading for upgrading. If you are interested, you can refer to them.

question 5: what preparations do you need to make to switch from functional testing to performance testing? This problem is relatively big. First, make clear a direction, whether it is client performance or server performance. Of course, it is usually server performance. After all, the server is the main body. Once everyone dies, the client can generally use TestBird's functional testing platform to solve it.

the performance of the server is a little more complicated. The first step is the design and scheme. Now, LR and jmeter are widely used tools, and ab is recommended by lazy tools. The realization of the performance of the server is to put pressure on the server by contracting out a large number of contracts, which is actually the quantitative behavior of network interface contracting out.

Secondly, we will encounter monitoring problems. Monitoring can be divided into script transactions, right and wrong monitoring, and general script code implementation. server monitor can be divided into system-level monitoring, application-level monitoring and database internal monitoring.

third, you may find that blind pressure monitoring can only find out obvious problems. At this time, you can go back and think about the scheme that should be considered at first, which is combined with your internal architecture, logic and business needs of the system to find out what is worth putting pressure on. This is just like writing a use case. Find the target at the beginning, but the performance result may be unknown. Of course, it is best to have a target or estimate the approximate behavior effect produced by a certain amount of data and online users. Suggestions are made.

question 6: what is the difficulty of performance testing? I haven't done it before, and I don't know much about it. Recently, the company will do app testing. Please ask? The difficulty of performance testing lies in the understanding of the system under test and the analysis of the test points. In order to realize the idea of testing, there are many ways, the means are always auxiliary, and only the idea is fundamental. Tools are not equal to performance testing. Don't think that you can understand performance testing by using LR. It is only the lowest level of test execution. Don't think that you can understand the performance test just by adjusting a few parameters, which is also a relatively low level.

tuning technology is not the main purpose of performance testing, and good performance is not tuned. Testers must understand the value of their existence, and the so-called "technology" is just some means to achieve their own testing purposes.

how to prove the validity of the test results is actually a very difficult problem, and it is worth taking time to think carefully. This process involves some very important contents, such as the establishment of user model, which will be described in detail later.

Performance testing is a process that needs continuous improvement. You just need to do better and do a little more things that you didn't think of before. After continuous accumulation, you will find that you have a deeper understanding of performance testing.

TestBird-an automated test platform for mobile games and apps

Question 7: Why do bumpers need to be tested for performance? Bumpers bear the risk of impact and need to mitigate the impact, so there are requirements on materials and structure

Question 8: What should I do to test the performance of computer hardware? A newly purchased computer or a newly upgraded computer needs to be tested with some heavy load, so that it can be found out whether the hardware is overclocked and whether the compatibility, stability and operating efficiency of the whole system are satisfactory as soon as possible, and even if there is no problem with the hardware in the end, it can provide a "theoretical basis" for optimizing the computer and "squeezing out" the computer potential.

First, memory test

Most computers are unstable, and memory is one of the important reasons. The quality of memory is directly related to the overall performance and stability of the system. Usually, we judge the quality of memory, either by observing the memory work with the naked eye or by waiting for the memory to fail.

to test memory, we usually use MemTest. MemTest can not only thoroughly detect the stability of the memory, but also test the memory storage capacity and data retrieval capacity of the memory at the same time, so that users can know whether the memory currently in use on the machine can be trusted, and at the same time, it is also a portable application, which can be run by directly clicking the execution file.

the use of memtest is very simple. Opening MemTest will bring up the main interface. Users can fill in the memory capacity they want to test in the space. If they don't fill in it, the software defaults to "All Unused Memory", and then click the "Start Test" button below to start the test. In the test, just click "Stop Test" and the test can be terminated.

MemTest will cycle through the memory until the user terminates it. If there is any quality problem in the memory, MemTest will give a prompt. The longer this test is carried out, the higher the test of memory quality will be. Usually, as long as the following values are displayed after 2%, there is basically no quality problem in the memory.

Second, CPU stability test

Needless to say, the importance of this test. Usually, U instability is caused by temperature, such as excessive temperature or poor heat dissipation. If the problem is not found early, it will even be burned in serious cases. Even if it is not burned, U will run at high temperature for a long time, and "electron migration" will occur, which will affect its life.

to measure u, I usually use Hot CPU Tester and super pi.

the function of the hot CPU tester itself is to test the stability of the CPU and the advantages and disadvantages of the cooling system (temperature monitoring software is needed), but there is an additional function when copying the machine, that is, it can be found whether the CPU is overclocked.

after Hot CPU Tester is installed, a shortcut will be displayed on the desktop. double-click the shortcut to enter the main interface of Hot CPU Tester. On the left is the menu, where we can see seven Options, namely, Diagnotisc, burn-in, Benchmark, System Info, option, LiveSupport and abo.

stability testing mainly relies on two functions: "Diagnotisc" and "burn-in". Just click the "Run Test" button in the "Diagnotisc Test" window to start testing the system. This test is to let the computer continuously calculate various items, so as to keep the CPU utilization rate at 1% for a period of time, so as to test the stability of the CPU.

The middle window is the test items, including hard disk, memory, MMX, etc. The CPU resources are occupied below. Usually, the CPU utilization rate will reach 1% in less than a few minutes. If the player can make the Hot CPU Tester run for more than 1 hour in this state, then the stability of the CPU is up to the standard. In addition, because the CPU works at 1% for a long time, it will generate great heat. At this time, players can see the temperature of the CPU and the system through some temperature monitoring software, and judge the quality of the cooling system.

at the bottom right, the Hot CPU Tester will display the time when the test was conducted, and the time when the test started and ended. In addition, in the Option, you can also set some tests, such as the test time and the state of the hardware during the test. Some functions need to be registered before users can use them.

if the player only wants to test the CPU and memory, he can enter ... > >

question 9: how to test the performance? why should we consider computing power, memory, cache, computing scale, web system or server?

question 1: given the test requirements, how to use Loadrunner to do performance test? Write a script first, then take 1 users for concurrent test. After the test is completed, generate a result, in which you can see the actual number of concurrent users and the average response time, and then compare it with 1 and 3~5 seconds to see if it meets the requirements. It's very simple, a rough process. You can refer to it first.