Detailed explanation of pipeline pipeline mechanism in redis

Hello everyone , I meet you again , I'm the king of the whole stack .

One 、pipeline The background ：

redis The client executes a command 4 A process ：

 dispatch orders －〉 Order to line up －〉 Command execution －〉 Return results 

Copy

This process is called Round trip time( abbreviation RTT, Round trip time ),mget mset It effectively saves RTT, But most orders （ Such as hgetall, did not mhgetall） Batch operation is not supported , Need to consume N Time RTT , At this time, we need pipeline To solve this problem .

Two 、pepeline Performance of

1、 not used pipeline perform N Bar command

2、 Used pipeline perform N Bar command

3、 Performance comparison between the two

Summary ： This is the data from a set of statistical data , Use Pipeline The execution speed is faster than one by one , In particular, the greater the network delay between the client and the server , The more performance, the more physical .

3、 ... and 、 Native batch command (mset, mget) And Pipeline contrast

1、 Native batch commands are atomic ,pipeline It's nonatomic

( The concept of atomicity : A transaction is an indivisible minimum unit of work , Both succeed or fail . Atomic operation means that one of your business logic must be indivisible . Deal with one thing or succeed , Or they all failed , An atom is indivisible )

2、 Native batch command: one command is more than one key, but pipeline Support multiple commands （ There is a business ）, Non atomicity

3、 The native batch command is implemented on the server side , and pipeline It needs to be done by both the server and the client

Four 、Pipeline Use it correctly

In the following code , I used an array of user names , Of array elements key The value is corresponding to the user id, Once the user changes his user name , I will modify two redis value ：

• The number of user name changes of the current user needs +1
• Update the user name data corresponding to the current user

$redis = new Redis();

// Turn on pipeline mode , The representative temporarily puts the operation command in the pipeline 
$pipe = $redis->multi(Redis::PIPELINE);

// Loop through the data , Perform the operation 
foreach ($users as $user_id => $username)
{
    // User name modification times +1
    $pipe->incr('changes:' . $user_id);

    //  Update user name 
    $pipe->set('user:' . $user_id . ':username', $username);
}

// Start executing all commands in the pipeline 
$pipe->exec();

Copy

What operation is put in the pipe , There are no restrictions , Even if you put the operation of obtaining data ok Of . Now suppose we want to give someone redis key value +1, But get another redis key It's worth it value data . The following code is an operation to update the number of times a user is accessed , Another operation is to obtain user information data .

$redis = new Redis();

// Turn on pipeline mode 
$pipe = $redis->multi(Redis::PIPELINE);

// Loop through the data , Perform the operation 
foreach ($users as $user_id => $username)
{
    //  The number of times the user was visited +1
    $pipe->incr('accessed:' . $user_id);

    //  Get user data records 
    $pipe->get('user:' . $user_id);
}
//  Start executing all commands in the pipeline 
$users = $pipe->exec();

//  Print data 
print_r($users);

Copy

Be careful , Because every command in the pipeline will return data , So the final printed array , Will contain incr Records brought by operation , There are also those pulled down from getting user operations redis key Value is used as the index value of the print array .

But there is one advantage , The data returned by each operation command in the pipeline is stored in sequence in the pipeline ,key The value is 0,1,2 such . What data do we want , Just deal with it a little by yourself .

If we like to cancel the pipeline operation , Use the following code ：

$pipe->discard();

Copy

summary ：pipeline Although easy to use , But every time pipeline The number of commands per assembly cannot be unlimited , Otherwise, one assembly pipeline Too much data , On the one hand, it will increase the waiting time of the client , On the other hand, it will cause some network congestion , It's possible to put a pipeline Split into several smaller pipeline To complete .

Publisher ： Full stack programmer stack length , Reprint please indicate the source ：https://javaforall.cn/111449.html Link to the original text ：https://javaforall.cn