Shulou(Shulou.com)05/31 Report--

This article mainly explains "What are the common problems in Web performance testing", interested friends may wish to take a look. The method introduced in this paper is simple, fast and practical. Let Xiaobian take you to learn "What are the common problems in Web performance testing"!

1, is currently studying jmeter, ask the initial performance, should start from what point?

Jmeter and LR are currently the preferred tools for performance testing. If you have a LR background, Jmeter will get started quickly. Focus on understanding the concepts of threads and processes. Processes and LR are similar. If not, from the point of view of entry, I suggest to understand the common network protocols and operating system threads and processes, Jmeter is ideal for java, you can consider understanding some programming basics. PS: Its practical Jmeter learning interface tests first will better understand performance.

2. What do you need to prepare for the zero-based, computer hardware professional transition to this industry?

Computer hardware already has the basic computer knowledge, it is relatively easy to transfer software, you can consider preparing from the following dimensions:

1)First prepare your own network and operating system. The network is mainly the application layer protocol, paving the way for performance and interface testing. The operating system is mainly to build a test environment.

2)It is recommended to be familiar with a programming language, java or Python are preferred languages, do not need to be proficient, at least to be able to write simple scripts

3)Find a comprehensive introduction to professional software testing books intensive reading, must be intensive reading.

4)If possible, it's best to practice on projects, starting with functional tests

3. At present, when doing interface test, I will still be at a loss when I advance. I hope you can give some suggestions. In addition, when you do management, what attitude or attitude do you generally have towards employees? How to cultivate, how to get along, how to assign tasks, control project progress, quality?

This is a big problem. Let me pick some highlights:

1)Interface testing is actually protocol testing. It is recommended that you start with network protocols so that you can do interface testing well.

2)As for the test management work, I think it varies from person to person. Some people are naturally strong, but some are peaceful. The most important thing is their own strength. If they are not strong enough, a saying goes,"A nest of bears is this truth. Attitude is not important, entity is important."

3)As for training, getting along, and assigning tasks, these are the management rules. The general idea is to make up for their shortcomings with their strengths. No one is perfect. Moreover, most of them will feel that they are better than the leaders. It is more important to let each employee give full play to their value and have a sense of achievement.

4)Controlling project progress and quality is a matter of method. There are many methods to control it, such as test version control, defect analysis, etc. You can refer to software engineering, agile process and other aspects of information. want to help you

4, with Loadrunner to do pressure test, response time is much higher than the actual, with LR to do pressure test, the average response time is tens of seconds, the actual manual open link less than 1 second, the gap is very large, what causes this problem?

For example, in the picture below, there are only a dozen vusers, and the response time is ten seconds, but the actual experience is still very fast. How to solve it?

Initial guess should be caused by improper response time settings, for example: you want to record the login script, and record the login response time, you insert a transaction function in the LR script (note that it is a group of functions), the response time is 5s, but the actual login you feel less than 1s, then it may be that your function is not located correctly, including the time to enter the username and password, or your response time function contains thinking time. In the former case, you need to adjust the position of the transaction; in the latter case, you need to exclude the thinking time from the run time setting. for reference

5. Ask how to grab the page elements of the program? Grasping page elements like Selenium web? Is there any good tool or method?

If you just want to grab page elements, I recommend Chrome Developer tools, which you can see under Elements. For automated testing, QTP (ALM) can also be considered in addition to selenium.

6. How should software fatigue test be done?

Generally, pressure test is done. Pressure test varies according to different projects and businesses. Three kinds of pressure test are recommended. One is 80% of 4H peak pressure; one is 24H, pressure force can be controlled at 60% 70%; and the other is 7*24 hours (some are continuous 50% pressure, and some are tested according to time fluctuation pressure value).

7. What preparations need to be made before stress testing B/S systems? How to objectively analyze a system? Pressure testing tools have only been exposed to Loadrunner before, so I intend to use Loadrunner, not too familiar with it, the system indicators that need to be set inside look very troublesome, I do not know how to analyze the system. And what else do I have to write about before I do the stress test besides choosing the familiar tool?

Your question is difficult to explain in one or two sentences. For the preliminary preparation work, I can only talk about my humble opinion in the performance test project:

1) Conduct a preliminary analysis of requirements to determine which aspects need performance testing, that is, where is the most stressful point for your system

2) Look at the existing resources, prepare the environment in advance, and it is best to test the environment and the production environment 1:1 (it is really impossible to do it as close as possible, this is very important, otherwise you have to do equal conversion)

3) Look at the protocols used by the system under test and various configurations such as operating system and application server, and select matching test tools (most WEB systems LR can cope)

4) Look at the time. In most cases, the time for performance testing is not sufficient, so it is necessary to focus on priority testing.

8. How to choose between Loadrunner and Jmeter?

All depends on the mood, joke ha ~~ if from the point of view of learning entry to see the code and network foundation, if the code and network foundation is good directly with Jmeter entry is good, conversely with LR entry is better. If you look at which is more appropriate from the perspective of enterprise applications, it is better for the system under test.

Tools are just formality, understanding the fundamentals of performance testing can be done with any tool.

9. How to locate the problem with Java frequent GC?

Try using profiler to look for memory exceptions, such as too many object creations for a short time, or large object creations.

10, I want to achieve 50 users concurrent work card check-in, parameterization, iteration has been set, and then, in front of the check-in function added a collection point function, running the results found that the user check-in time returned is one minute, and did not sign in at the same time point! Ask why? How to solve it?

How about we remove the rendezvous point first? Start 50 users in parallel at the same time (concurrent operation can be achieved without setting the assembly point). If it lasts for one minute, please check the transaction time and thinking time. The transaction time needs to be manually configured. The thinking time defaults to the startup state. You can check it in Runtime Settings. It is possible that the 1 minute you finally get is the time for the entire script to run once or the result including the thinking time, not the time for punching at the same time.

11. How are e-commerce spike products tested?

It is not much different from the tests of other products, mainly because the requirements for time points are relatively high. You can consider using assembly point functions in performance test scripts to achieve concurrency in the same second.

12. Is stress testing the same as performance testing?

The classification methods are different and inconclusive. Generally speaking, performance test is a general term for this kind of test. I prefer the following classification

Performance test (narrow sense)-Performance test method is to test whether the performance of the system meets the production performance requirements by simulating the combination of business stress and use scenarios of production operation under specific operating environment.

Benchmark testing-Simulate a certain number of users running one or more services in a certain software, hardware and network environment, and use the test results as baseline data for subsequent testing activities.

Load testing-by applying pressure to the system under test until a performance metric reaches a limit, such as "response time" exceeding a predetermined metric or a resource has reached saturation.

Stress testing-stress testing, also known as stress testing, mainly tests the ability of the system to handle sessions under certain saturation conditions, such as cpu, memory saturation, and whether the system will have errors. PS: Continuous operation for a period of time (2~4 hours) at about 80% of the limit value

Configuration test-configuration test method through the adjustment of the software\hardware environment of the system under test, to understand the extent of different impact on the system performance, so as to find the optimal allocation principle of system resources.

Reliability test--Under the condition that the system is loaded with certain service pressure, the system is operated for a period of time to check whether the system is stable.

Concurrent testing-Concurrent testing methods simulate concurrent user access to test whether there are deadlocks or other performance problems when multiple users access the same application, the same module, or data records concurrently.

PS: Classification is actually not so important. In actual projects, it is often mixed application.

13. Web performance testing In addition to concurrent login, what are the more common test scenarios?

Scenario depends on the business, for example, you are an e-commerce website, you must test the situation of placing orders at the same time? ?; If you are a hospital registration website, you must test multiple people to grab a doctor's number source; if you are a bank system, consider multiple people withdrawing money at the same time?

How many millions of users does the general website pressure?

This depends on the calculation method, the theoretical value of 1 million/day of business access to the split per second is very small; but this does not conform to the actual situation.

If the system is online, you can directly take the online data as a reference. There are several peaks in a day, and the distribution of peak periods can calculate the number of concurrent users.

If the system is not online, the 2/8 principle can be used to derive the service access volume, where 80% of users are concentrated in 20% of the time.

15, currently encountered a problem, in a configuration of 8g, i3 win7 system running stress test, concurrency always does not go up, always around 140 will appear abnormal, timeout and other problems, how to analyze it, is it because of the server or computer problems?

Server stress can be analyzed from the following dimensions:

Is there any restriction on network traffic?

Whether database/application server reports exception, if yes, check log;

Check the resource monitoring status of the operating system, how the CPU utilization rate is, and whether it has reached 100%

Whether there are performance problems in the code, you can manually access the system during large concurrent access to see if there are exceptions in the business.

At this point, I believe that everyone has a deeper understanding of "what are the common problems in Web performance testing". Let's actually operate it! Here is the website, more related content can enter the relevant channels for inquiry, pay attention to us, continue to learn!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.