How to implement Redis read and write separation in go

Why do we need to understand the RESP protocol?

Regarding this issue, I would like to explain through an example why we need to understand the RESP protocol when we write Redis middleware.

The above code is a very simple TCP server. We listen to the 8888 port and try to use redis -cli -p 8888After connecting to the server, then view the printed application layer messages.

We try to execute this code and enter redis-cli -p 8888 to connect.

The server we wrote gets the message from the redis client:

*1
$7
COMMAND

The above is the content of the RESP protocol, so if we want to write a Redis middleware, we need to first understand the RESP protocol. .

What is RESP protocol

The official website has relevant explanations: https://redis.io/docs/reference/protocol-spec/

RESPThe protocol was originally designed for communication between Redis servers and clients. The protocol was introduced in Redis 1.2, and in Redis 2.0, Become the standard protocol for Redis communication. This protocol has the following advantages:

• Simple implementation

• Quick analysis

• Directly readable

RESPAccording to its protocol prefix, different data types can be serialized, such as integers, strings, arrays, etc., and normal output and error output can also be marked. . In addition to pipelines and publish and subscribe, the RESP protocol should be the simplest request-response protocol. For more introduction, you can check out the official documentation commented above.

RESP protocol specification

RESPDifferent parts of the protocol are separated using \r\n (line feed character), which supports 5 types of data The types are: simple string, error, integer, complex string and array. Let’s make a table to explain.

##Complex string$Complex string starts with $Array*Array starts with *

我当初看到这个的时候，也是迷迷糊糊的，到底什么意思呢？ 哎，我们举个例子你就明白了。

若我们想执行

set juejinName pdudo

若使用RESP 协议应当如何编写呢？应当编写如下：

*3
$3
set
$10
juejinName
$5
pdudo

那我们来解释一下*3代表有3个数组，而$3代表复杂字符串有长度为3，值为set， $10代表复杂字符串长度为10，值为juejinName， $5代表复杂字符串长度为5，值为pdudo。

我们结合上述信息，可以画一张图。

这就是协议的内容了。

而协议前缀+、-、:则要简单的多，直接跟数据即可，

例如:

+

+OK

-

-ERR syntax error

:

:3

如何使用该协议请求Redis

我们已经学习了相关的RESP协议，那么我们如何学习呢？ 我们可以使用telnet命令来操作即可。

在此，我们准备几条命令，我们会将其转换为RESP格式，且将其发送到redis服务器。

命令

set name pdudo
get name
lpush pn 1
llen pn

转换为RESP格式

*3
$3
set
$4
name
$5
pdudo
*2
$3
get
$4
name
*3
$5
lpush
$2
pn
$1
1
*2
$4
llen
$2
pn

我们将其放置到telnet中执行一下呢

现在回头看看官网文档所提及的，该协议实现简单，直接可阅读，是不是理解的更加深刻了呢？

使用go编写Redis中间件实现读写分离

本篇暂不解释代码，而后单独开一篇谈论中间件代码。

实现该功能，其实本质上是区分命令是查询还是写入，若是查询，则直接转发到从库，而写入，则转发到主库即可，其架构图可以理解为如下：

我们已经有了目前的架构。

Type	Prefix	Remarks
Simple string		Simple string starts with
Error data	-	Error data starts with -
Integer	:	Integer starts with:

主机	端口	密码	角色
127.0.0.1	6379	无	主库
127.0.0.1	7380	无	从库

The above is the detailed content of How to implement Redis read and write separation in go. For more information, please follow other related articles on the PHP Chinese website!