Redis realizes the spike function

In real life, the spike function is more common, such as 12306 ticket grabbing, e-commerce system spike activities, etc. The so-called spike, from the perspective of application business, means that multiple users "compete" for a certain resource in a short period of time. The resource here is a commodity in most spike scenarios; from a technical point of view, it is a combination of multiple threads. Resources to operate. Therefore, in order to realize the spike function, it is necessary to control the thread's contention for resources, not only to ensure high efficiency and concurrency, but also to ensure the correctness of the operation and meet the actual business needs.

The optimization ideas for spike are:

• Write to memory.
• Realize multi-threaded asynchronous processing.
• Realize distributed processing.

[Example] An example of using multi-threading to achieve 1000 people killing 100 mobile phones in seconds.

(1) In the pom.xml configuration information file, add the dependencies of the Redis launcher and Jedis client:

<!-- Redis启动器 -->
<dependency>
    <groupId>org.springframework.boot</groupId>
    <artifactId>spring-boot-starter-data-redis</artifactId>
    <version>2.4.0</version>
</dependency>

<!-- Jedis客户端依赖 -->
<dependency>
    <groupId>redis.clients</groupId>
    <artifactId>jedis</artifactId>
    <version>3.3.0</version>
</dependency>

(2) SecondKill implementation (SecondKill.java): Create multiple threads and use the transaction function of Redis to realize the second kill function.

package com.pjb.seckill;

import redis.clients.jedis.Jedis;
import redis.clients.jedis.Transaction;
import java.util.List;

/**
 * 秒杀抢购
 * @author pan_junbiao
 **/
public class SecondKill implements Runnable
{
    String key = "iphone";
    Jedis jedis = new Jedis("127.0.0.1",6379);
    String userInfo;

    public SecondKill(String userInfo)
    {
        this.userInfo = userInfo;
    }

    @Override
    public void run()
    {
        try
        {
            jedis.watch(key); //watchkeys

            String val = jedis.get(key);
            int valint = Integer.valueOf(val);

            if(valint<=100 && valint>=1)
            {
                //1、使用MULTI命令开启事务
                Transaction tx = jedis.multi();

                //2、事务命令入队
                tx.incrBy(key,-1);

                //3、使用EXEC命令执行事务
                //提交事务。如果此时watchkeys被改动了，则返回null
                List<Object> list = tx.exec();

                if(list==null || list.size()==0)
                {
                    String failuserinfo = "fail_" + userInfo;
                    String failinfo = "用户：" + failuserinfo + "商品争抢失败，抢购失败";
                    System.out.println(failinfo);
                    //抢购失败业务逻辑
                    jedis.setnx(failuserinfo,failinfo);
                }
                else
                {
                    for(Object succ : list)
                    {
                        String succuserinfo = "succ_" + succ.toString() + "_" + userInfo;
                        String succinfo = "用户：" + succuserinfo + " 抢购成功，当前抢购成功人数：" + (1 -(valint -100));
                        System.out.println(succinfo);
                        //抢购成功业务逻辑
                        jedis.setnx(succuserinfo,succinfo);
                    }
                }
            }
            else
            {
                String failuserinfo = "kcfail_" + userInfo;
                String failinfo1 = "用户：" + failuserinfo + " 商品被抢购完毕，抢购失败";
                System.out.println(failinfo1);
                jedis.setnx(failuserinfo,failinfo1);
                return;
            }
        }
        catch(Exception ex)
        {
            ex.printStackTrace();
        }
        finally
        {
            jedis.close();
        }
    }
}

(3) The main program (SecondKillTest.java): the entry program of the spike function, used to create a thread pool, at the same time, generate a user ID, call the spike function code, and complete the spike task.

package com.pjb.seckill;

import com.pjb.util.RedisHelper;
import redis.clients.jedis.Jedis;

import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;

/**
 * Redis秒杀功能的实现，1000人抢购100部手机
 * @author pan_junbiao
 **/
public class SecondKillTest
{
    public static void main(String[] args)
    {
        final String key = "iphone";
        //20个线程池并发数
        ExecutorService executor = Executors.newFixedThreadPool(20);

        final Jedis jedis = new Jedis("127.0.0.1",6379);
        jedis.del(key); //先删除
        jedis.set(key,"100"); //设置起始的抢购数
        jedis.close();

        for(int i=0; i<1000; i++)
        {
            String userInfo = "pan_junbiao的博客" + i;
            executor.execute(new SecondKill(userInfo));
        }
        executor.shutdown();
    }
}

Results of the: