PHP and Redis implement rush buying and flash sales under high concurrency

This article mainly shares with you PHP and Redis to implement rush buying and flash sales under high concurrency. Rush buying and flash sales are very common scenarios. During interviews, interviewers often ask, for example, asking you what is the flash sale on Taobao? How to implement it, etc., I hope it can help everyone

It is very simple to implement flash sales and flash sales, but there are some problems that need to be solved, mainly focusing on two problems:

1. High concurrency pairing The pressure generated by the database
2. How to solve the correct inventory reduction ("oversold" problem) under competitive conditions

The first problem is very simple for PHP. The database can be alleviated by caching technology Pressure, such as memcache, redis and other caching technologies.
The second question is more complicated:

Conventional writing:
Query the inventory of the corresponding product to see if it is greater than 0, and then perform operations such as generating an order, but when judging whether the inventory is greater than 0 If there is high concurrency, there will be problems, resulting in negative inventory.

<?php
$conn=mysql_connect("localhost","big","123456");
if(!$conn){
echo "connect failed";
exit;
}
mysql_select_db("big",$conn);
mysql_query("set names utf8");
$price=10;
$user_id=1;
$goods_id=1;
$sku_id=11;
$number=1;
//生成唯一订单
function build_order_no(){
return date(&#39;ymd&#39;).substr(implode(NULL, array_map(&#39;ord&#39;, str_split(substr(uniqid(), 7, 13), 1))), 0, 8);
}
//记录日志
function insertLog($event,$type=0){
global $conn;
$sql="insert into ih_log(event,type)
values(&#39;$event&#39;,&#39;$type&#39;)";
mysql_query($sql,$conn);
}
//模拟下单操作
//库存是否大于0
$sql="select number from ih_store where goods_id=&#39;$goods_id&#39; and sku_id=&#39;$sku_id&#39;";
//解锁 此时ih_store数据中goods_id=&#39;$goods_id&#39; and sku_id=&#39;$sku_id&#39; 的数据被锁住(注3)，其它事务必须等待此次事务 提交后才能执行
$rs=mysql_query($sql,$conn);
$row=mysql_fetch_assoc($rs);
if($row[&#39;number&#39;]>0){//高并发下会导致超卖
$order_sn=build_order_no();
//生成订单
$sql="insert into ih_order(order_sn,user_id,goods_id,sku_id,price)
values(&#39;$order_sn&#39;,&#39;$user_id&#39;,&#39;$goods_id&#39;,&#39;$sku_id&#39;,&#39;$price&#39;)";
$order_rs=mysql_query($sql,$conn);
//库存减少
$sql="update ih_store set number=number-{$number} where sku_id=&#39;$sku_id&#39;";
$store_rs=mysql_query($sql,$conn);
if(mysql_affected_rows()){
insertLog(&#39;库存减少成功&#39;);
}else{
insertLog(&#39;库存减少失败&#39;);
}
}else{
insertLog(&#39;库存不够&#39;);
}
出现这种情况怎么办呢？来看几种优化方法：
优化方案1：将库存字段number字段设为unsigned，当库存为0时，因为字段不能为负数，将会返回false
//库存减少
$sql="update ih_store set number=number-{$number} where sku_id=&#39;$sku_id&#39; and number>0";
$store_rs=mysql_query($sql,$conn);
if(mysql_affected_rows()){
insertLog(&#39;库存减少成功&#39;);
}
优化方案2：使用MySQL的事务，锁住操作的行
<?php
$conn=mysql_connect("localhost","big","123456");
if(!$conn){
echo "connect failed";
exit;
}
mysql_select_db("big",$conn);
mysql_query("set names utf8");
$price=10;
$user_id=1;
$goods_id=1;
$sku_id=11;
$number=1;
//生成唯一订单号
function build_order_no(){
return date(&#39;ymd&#39;).substr(implode(NULL, array_map(&#39;ord&#39;, str_split(substr(uniqid(), 7, 13), 1))), 0, 8);
}
//记录日志
function insertLog($event,$type=0){
global $conn;
$sql="insert into ih_log(event,type)
values(&#39;$event&#39;,&#39;$type&#39;)";
mysql_query($sql,$conn);
}
//模拟下单操作
//库存是否大于0
mysql_query("BEGIN"); //开始事务
$sql="select number from ih_store where goods_id=&#39;$goods_id&#39; and sku_id=&#39;$sku_id&#39; FOR UPDATE";//此时这条记录被锁住,其它事务必须等待此次事务提交后才能执行
$rs=mysql_query($sql,$conn);
$row=mysql_fetch_assoc($rs);
if($row[&#39;number&#39;]>0){
//生成订单
$order_sn=build_order_no();
$sql="insert into ih_order(order_sn,user_id,goods_id,sku_id,price)
values(&#39;$order_sn&#39;,&#39;$user_id&#39;,&#39;$goods_id&#39;,&#39;$sku_id&#39;,&#39;$price&#39;)";
$order_rs=mysql_query($sql,$conn);
//库存减少
$sql="update ih_store set number=number-{$number} where sku_id=&#39;$sku_id&#39;";
$store_rs=mysql_query($sql,$conn);
if(mysql_affected_rows()){
insertLog(&#39;库存减少成功&#39;);
mysql_query("COMMIT");//事务提交即解锁
}else{
insertLog(&#39;库存减少失败&#39;);
}
}else{
insertLog(&#39;库存不够&#39;);
mysql_query("ROLLBACK");
}
优化方案3：使用非阻塞的文件排他锁
<?php
$conn=mysql_connect("localhost","root","123456");
if(!$conn){
echo "connect failed";
exit;
}
mysql_select_db("big-bak",$conn);
mysql_query("set names utf8");
$price=10;
$user_id=1;
$goods_id=1;
$sku_id=11;
$number=1;
//生成唯一订单号
function build_order_no(){
return date(&#39;ymd&#39;).substr(implode(NULL, array_map(&#39;ord&#39;, str_split(substr(uniqid(), 7, 13), 1))), 0, 8);
}
//记录日志
function insertLog($event,$type=0){
global $conn;
$sql="insert into ih_log(event,type)
values(&#39;$event&#39;,&#39;$type&#39;)";
mysql_query($sql,$conn);
}
$fp = fopen("lock.txt", "w+");
if(!flock($fp,LOCK_EX | LOCK_NB)){
echo "系统繁忙，请稍后再试";
return;
}
//下单
$sql="select number from ih_store where goods_id=&#39;$goods_id&#39; and sku_id=&#39;$sku_id&#39;";
$rs=mysql_query($sql,$conn);
$row=mysql_fetch_assoc($rs);
if($row[&#39;number&#39;]>0){//库存是否大于0
//模拟下单操作
$order_sn=build_order_no();
$sql="insert into ih_order(order_sn,user_id,goods_id,sku_id,price)
values(&#39;$order_sn&#39;,&#39;$user_id&#39;,&#39;$goods_id&#39;,&#39;$sku_id&#39;,&#39;$price&#39;)";
$order_rs=mysql_query($sql,$conn);
//库存减少
$sql="update ih_store set number=number-{$number} where sku_id=&#39;$sku_id&#39;";
$store_rs=mysql_query($sql,$conn);
if(mysql_affected_rows()){
insertLog(&#39;库存减少成功&#39;);
flock($fp,LOCK_UN);//释放锁
}else{
insertLog(&#39;库存减少失败&#39;);
}
}else{
insertLog(&#39;库存不够&#39;);
}
fclose($fp);
优化方案4：使用redis队列，因为pop操作是原子的，即使有很多用户同时到达，也是依次执行，推荐使用（mysql事务在高并发下性能下降很厉害，文件锁的方式也是）
先将商品库存如队列
<?php
$store=1000;
$redis=new Redis();
$result=$redis->connect(&#39;127.0.0.1&#39;,6379);
$res=$redis->llen(&#39;goods_store&#39;);
echo $res;
$count=$store-$res;
for($i=0;$i<$count;$i++){
$redis->lpush(&#39;goods_store&#39;,1);
}
echo $redis->llen(&#39;goods_store&#39;);
抢购、描述逻辑
<?php
$conn=mysql_connect("localhost","big","123456");
if(!$conn){
echo "connect failed";
exit;
}
mysql_select_db("big",$conn);
mysql_query("set names utf8");
$price=10;
$user_id=1;
$goods_id=1;
$sku_id=11;
$number=1;
//生成唯一订单号
function build_order_no(){
return date(&#39;ymd&#39;).substr(implode(NULL, array_map(&#39;ord&#39;, str_split(substr(uniqid(), 7, 13), 1))), 0, 8);
}
//记录日志
function insertLog($event,$type=0){
global $conn;
$sql="insert into ih_log(event,type)
values(&#39;$event&#39;,&#39;$type&#39;)";
mysql_query($sql,$conn);
}
//模拟下单操作
//下单前判断redis队列库存量
$redis=new Redis();
$result=$redis->connect(&#39;127.0.0.1&#39;,6379);
$count=$redis->lpop(&#39;goods_store&#39;);
if(!$count){
insertLog(&#39;error:no store redis&#39;);
return;
}
//生成订单
$order_sn=build_order_no();
$sql="insert into ih_order(order_sn,user_id,goods_id,sku_id,price)
values(&#39;$order_sn&#39;,&#39;$user_id&#39;,&#39;$goods_id&#39;,&#39;$sku_id&#39;,&#39;$price&#39;)";
$order_rs=mysql_query($sql,$conn);
//库存减少
$sql="update ih_store set number=number-{$number} where sku_id=&#39;$sku_id&#39;";
$store_rs=mysql_query($sql,$conn);
if(mysql_affected_rows()){
insertLog(&#39;库存减少成功&#39;);
}else{
insertLog(&#39;库存减少失败&#39;);
}

The above is just a simple simulation of rush buying under high concurrency. The real scenario is much more complicated than this. There are many things to pay attention to, such as making the rush buying page static and calling the interface through ajax.

The above will cause one user to grab multiple items. Idea:
Needs a queuing queue, a buying result queue and an inventory queue. In the case of high concurrency, first enter the user into the queue, use a thread loop to remove a user from the queue, and determine whether the user is already in the rush-buying result queue. If it is, it has been snapped up, otherwise it is not snapped up, the inventory is reduced by 1, and the database is written. Put the user into the result queue.
When I was working on a shopping mall project, I used redis directly for flash sales. During this period, I looked at the above methods. Although they are different, they all achieve the same purpose. You can choose for yourself. Just be happy.

Related recommendations:

php+redis implements rush purchase function

How does php handle high concurrent requests for rush purchase functions

PHP implements concurrent snap-up function through locking

The above is the detailed content of PHP and Redis implement rush buying and flash sales under high concurrency. For more information, please follow other related articles on the PHP Chinese website!