Shulou(Shulou.com)06/03 Report--

Many friends often struggle with online and offline test results... So let's write an article like this

In fact, this problem itself does not need to be entangled, just like no matter how awesome twins are, they still have their differences. Performance testing itself is a means of estimating risks, troubleshooting bottlenecks, and understanding the current performance of the system. Just like when you were a child, you were a good child, but that doesn't mean you'll grow up to be a good child, maybe you'll be like Brother Haibo... Performance testing is just a means to minimize risk.

Moreover, the online and offline test results themselves are not very comparable, because:

1. Differences between offline and online machine environment configurations

2. The difference between offline and online business data. Although we want to simulate the user behavior to the maximum extent offline, you can't guarantee 100% simulation. So many users can you take into account???

3. The difference between offline and online pressure generation time, offline is to simulate the situation of high pressure and large complications, while online pressure is usually not large, and large complications are mainly concentrated in a few special periods.

Speaking of this, there will be children's children continue to struggle, then why do you still do tests, all inaccurate, do a hair???? Well, then I want to ask you a question. Different people drive a car and their driving habits are different, which will affect the car to different degrees. Since we can't test the simulation 100%, we might as well produce the car and sell it directly to you. It's too much effort to do some tests and road tests. Right? If you quit at this time, it would be dangerous. What if there was a big problem? Wouldn't it kill me? Uh... Do you understand now? Then you don't understand when you switch to performance testing?

The reason we do performance testing is simple:

1. Prevent and assess risks. If there are big problems, they can be found early and reduce risks. Here understanding is extremely simple, your program has the problem of memory leakage, is offline 2g and online 4g this memory difference will not have memory leakage???????????????? This is like, you can't ride a permanent bicycle, do you want to give you a strong card (made up...) You can ride a bicycle in an instant?

2. Front-end performance test. The front-end performance test can ensure the page performance and bring better user experience to users.

3. Single interface performance optimization. The main purpose is to optimize interface performance, troubleshoot interface performance problems, and application memory hazards.

For example, we will prepare several business scenarios, such as all-go DB and all-go cache. Under these scenarios, under the condition of applying the best processing capacity, we will check whether there is any performance improvement in the test, and the memory leakage caused by code problems.

4. Capacity assessment. It can be estimated according to the proportion of online machines and offline simulation.

Well, that should be all... Doesn't that seem understandable?

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.