A complete tutorial for getting started with redis: Pipeline

3.3.1　Pipeline Concept
Redis The client execution of a command is divided into the following four processes ：
1） dispatch orders
2） Order to line up
3） Command execution
4） Return results
among 1）+4） be called Round Trip Time（RTT, Round trip time ）.
Redis Batch operation commands are provided （ for example mget、mset etc. ）, Save effectively RTT. but
Most commands do not support batch operations , For example, to perform n Time hgetall command , did not
mhgetall Command exists , Need to consume n Time RTT.Redis The client and server of may be deployed in different places
On the same machine . For example, the client is in Beijing ,Redis The server is in Shanghai , The straight-line distance between the two places is about
1300 km , that 1 Time RTT Time =1300×2/（300000×2/3）=13 millisecond （ Light in a vacuum
The transmission speed is per second 30 Thousands of kilometers , Let's say the fiber is at the speed of light 2/3）, So the client is 1 second
Can only be carried out in about 80 The next order or so , This and Redis The high concurrency and high throughput of
Chi .
Pipeline（ Assembly line ） The mechanism can improve the above problems , It can combine a group of Redis Order to enter
Row assembly , Through a RTT Transferred to the Redis, Will this group again Redis The execution results of commands are in order
Return to the client , chart 3-5 Is not used Pipeline Yes n Bar command , The whole process needs n Time
RTT.

  chart 3-6 For the use of Pipeline Yes n subcommand , The whole process needs 1 Time RTT.
Pipeline It's not a new technology or mechanism , Many technologies have been used . and RTT
In different network environments, there will be different , For example, the machine room and the same machine will be faster , Cross machine room and cross region
It will be slow .Redis The actual execution time of the command is usually at the microsecond level , That's why there is Redis sex
The bottleneck is the network .
redis-cli Of --pipe Option is actually using Pipeline Mechanism , For example, the following operation will set
hello world and incr counter Two commands assemble ：
echo -en '*3\r\n$3\r\nSET\r\n$5\r\nhello\r\n$5\r\nworld\r\n*2\r\n$4\r\nincr\r\
n$7\r\ncounter\r\n' | redis-cli --pipe

But most developers prefer to use... In high-level language clients Pipeline, At present, the
part Redis All clients support Pipeline, The first 4 In chapter, we will introduce how to pass Java Of Redis customer
Home end Jedis Use Pipeline function .

3.3.2　 Performance testing
surface 3-1 The non Pipeline and Pipeline perform 10000 Time set operation
The effect of , Two conclusions can be drawn as follows ：
·Pipeline Execution speed is generally faster than item by item execution .
· The greater the network delay between the client and the server ,Pipeline The more obvious the effect of .

Due to the different test environment, the specific number may be different , This test Pipeline Carry... Every time
belt 100 Bar command .

 3.3.3　 Native batch command and Pipeline contrast
have access to Pipeline Simulate the effect of batch operation , But when using it, we should pay attention to its relationship with the original
The difference between generating batch commands , It includes the following points ：
· The native batch command is atomic ,Pipeline It's non atomic .
· A native batch command is a command that corresponds to multiple key,Pipeline Supports multiple commands .
· The native batch command is Redis The server supports the implementation of , and Pipeline The server and the customer are required
The common implementation of the end .

3.3.4　 Best practices
Pipeline Although easy to use , But every time Pipeline The number of assembly commands cannot be unlimited , no
One time assembly Pipeline Too much data , On the one hand, it will increase the waiting time of the client , The other side
It can cause a certain amount of network congestion , It's possible to put a Pipeline Split into multiple
smaller Pipeline To complete .
Pipeline Only one can be operated Redis example , But even in distributed Redis Scene , Can also be
As an important optimization means of batch operation , For details, see page 11 Chapter .