What is the difference between WeChat development for 10 million fans and a few thousand fans?

(ps: I was tortured when I went to a big company for an interview. I went back and looked for information and found that there was very little in this area, and I was still confused.)

For example:
1. High concurrency and large traffic,
2. Disconnection and reconnection
3. The amount of data is too large and the request times out
4. Pushed text user synchronization and other issues.

I hope experienced experts can share with you how to develop in the case of large traffic. When developing with low user volume, you often don’t pay attention to the pits you dig and step on the thunder; and how to build large traffic samples for testing. , as well as learning directions and recommended books in this area.
(I am a back-end scumbag with 1 year of experience, thank you for your attention and answers) over!

Reply content:

(ps: I was tortured when I went to a big company for an interview. I went back and looked for information and found that there was very little in this area, and I was still confused.)

For example:
1. High concurrency and large traffic,
2. Disconnection and reconnection
3. The amount of data is too large and the request times out
4. Pushed text user synchronization and other issues.

I hope experienced experts can share with you how to develop in the case of large traffic. When developing with low user volume, you often don’t pay attention to the pits you dig and step on the thunder; and how to build large traffic samples for testing. , as well as learning directions and recommended books in this area.
(I am a back-end scumbag with 1 year of experience, thank you for your attention and answers) over!

Nowadays, interviewers like to ask these pretentious questions. Few of them know how to actually operate it. What they know is just theory. If you are not a technical director, you will not be given the opportunity to implement it. 10 interviewers asked this question 9 are show-offs.

Let me talk about what I have used
1 First of all, tens of millions of users depends on what kind of project you are. Ordinary projects do not have many 1,000,000-level users, because ordinary projects will not have user activity at all
2 Definition of high concurrency: same The number of IP requests per second. From my personal understanding, high concurrency refers to two aspects, one is your service, and the other is your program business logic.
Services: apache, nginx + fastcgi
Programming: I personally think this is the key point, because the bottleneck of web applications lies in the database. Your business logic will add, delete, check and modify the database. Therefore, when the current request volume reaches a certain base, it is likely to cause the memory occupied by the database to increase, and even cause your table to deadlock, or the database to crash.
To solve this problem: 1. First, you need to write a log at the entrance of your program to record the address and parameters of each request. 2 Analyze which requests have too high access frequency based on your logs (it is impossible for every request to access high frequency) 3 Find out the addresses with too high access requests, analyze the corresponding business logic, optimize the code and SQL statements, and do it according to functional requirements Proper caching
Database aspects: 1. Based on the above analysis results, tables that require frequent operations are obtained. Most of them are database query operations. (Generally, tables with frequent operations include: user information table. General tables with large amounts of data: financial record table, user log table.) Then divide the tables according to business logic. 1. Put the fields that are frequently queried in the user table into one table. 2. Put the user statistical information and unusual data into another table or tables. 3 For tables with large amounts of data, you can use time periods to divide tables, such as one financial record table and user log table for a month or a week. Another way is to divide tables according to business logic, such as one table for active user information and one table for inactive users. Another table is divided into tables according to user ID. For example, a table with the first digit of user ID being 1, a table with 2, and so on to create 10 tables.
Then use read-write separation to establish a master-slave library. There are many more things to say in a more detailed direction. But for general companies, there is really no need. If you have such a large number of users, it means that the company has money. If the company has money, many things can be improved through hardware, and it can hire full-time personnel at high prices. You are just interviewing PHP. I don’t have a log table and I haven’t done anything about the daily increase of 10 million.

3 Don’t be intimidated by this kind of people, just go to Baidu and memorize the answers, and memorize them for you next time you meet them.

Maybe their company is not hiring, but they just encountered such a problem and didn’t want to spend money, so they made it an interview question. . .