Thoughts on project development

Nearly period of time , Then reorganize your knowledge structure , I have some ideas about project concurrency and optimization , Take a brief note of ; You guys have good suggestions and ideas , Hope to share , If there are mistakes in the content , Hope to point out , thank ！

Highly concurrent projects are hard to reach , But programmers , We can create some concurrency by ourselves （ Although it's very low , Ha ha ha ）

I have written a small second kill before Demo, interested , You can look at it .

seckill Demo：https://github.com/Vmetrio/spike.git

For this project , I'm thinking about how to optimize , Solve the problem of a large number of visits ！

  One 、 Single machine performance optimization

①  Use Jmeter、wrk Isobaric tool , Of pressure measurement items QPS and TPS.

wrk：https://github.com/wg/wrk

QPS： full name Queries Per Second, intend “ Query rate per second ”, Is the number of queries a server can respond to per second , It is a measure of how much traffic a specific query server handles in a given period of time .

To put it simply ,QPS = req/sec = Number of requests / second . It represents the performance and maximum throughput of the server machine .

TPS namely Transactions Per Second Abbreviation , The number of transactions processed per second . A transaction is a process in which a client sends a request to the server and the server responds .

The client starts timing when it sends the request , End timing after receiving server response , This is used to calculate the time used and the number of transactions completed , The final assessment score made with this information .

TPS The process includes ： The client requests the server 、 Server internal processing 、 The server returns to the client .

QPS It's basically like Tps, But here's the difference , A visit to a page , To form a TPS; But a page request , There may be multiple requests to the server , The server requests for these , You can count in “QPS” In .

for example , Visit one Index The page will request the server 3 Time , Including once html, once css, once js, Then visiting this page will produce a “T”, Produce three “Q”.

② I'm on a real project , Use wrk During pressure measurement , Different modules of the project , Performance is different , In fact, each module has different concurrency indicators , have access to wrk, Write pressure test scripts for different modules .

③ Single machine optimization , Need to be right JVM tuning ; In the case of multithreading , Also consider whether the variables used are thread safe , for example ：HashMap The problem that leads to the dead cycle ,（ stay put During operation , When two threads are triggered at the same time rehash operation , A closed loop ）

④ JVM And code optimization , We will encounter database problems ,mysql The default maximum number of connections to the database is 100, More connections , It will make the database unavailable . therefore , The whole project needs to use load to solve new problems .

Two 、 Load balancing shares the access pressure

① When users visit , We can use Nginx To do the load balancing of the project （ polling 、ip_hash、 Rules such as weighting ）, Deploy the project to multiple servers .

② The project is behind multiple servers , Still haven't solved the problem of database , For many visits , Little change in data , We can use cache to solve , Use redis As cache middleware , At the same time, it can also solve the sharing among multiple projects session problem （ User information, etc ）.

③ Again using wrk Pressure test each module of the project , There will be new problems （ For example, server configuration 、 Different performance , The processing capacity is also different , You need to adjust the load algorithm ,Nginx Polling weighting of ）

3、 ... and 、 Use middleware for different modules

① Full text search elasticsearch; Message queue kafka; cache redis; Microservices springcloud; Load balancing nginx、zookeeper wait .

② Component clustering , To avoid a single point of failure , Cluster the components , To maintain high availability .

③ You need to learn and understand the ideas and underlying principles of various components , Use different technologies for different business scenarios .

④ Project degradation , Current limiting treatment, etc .

1、 Modules can process asynchronously , Using asynchronous processing

2、 Using caching technology

To be continued ......

QPS、TPS What is it? ：https://www.cnblogs.com/leadership/p/11555409.html