MySQL performance optimization (5): principle and implementation of master-slave synchronization

1 What is master-slave synchronization

        MySQL Master-slave synchronization means that data can be synchronized from one MySQL The database server master node is replicated to one or more slave nodes .MySQL Asynchronous replication is used by default , In this way, the slave node does not need to visit the master server all the time to update its data , Data can be updated on a remote connection , The slave node can copy all databases in the master database or specific databases , Or a specific watch .

2 Application scenarios

2.1 Read / write separation

         In the development work , Sometimes I meet someone sql Statement needs lock table , Resulting in temporary inability to use the read service , This will affect the existing business , Use master-slave replication , Let the main database be responsible for writing , Read from the library , such , Even if the main database appears to lock the table , The normal operation of the business can also be ensured by reading from the database . When the reading pressure increases , You can also configure multiple slave databases , Share the pressure of reading data together , Improve the stability of the system .

2.2 Architecture extension

         With the increase of business access in the system , If it is a stand-alone deployment database , It will lead to I/O Access frequency is too high . With master-slave replication , Add multiple data storage nodes , Distribute the load across multiple slave nodes , Reduce the cost of stand-alone disk I/O Frequency of visits , Improve the... Of a single machine I/O performance .

3 Schematic diagram

Operation principle after master-slave synchronization :

         The main database is generally only responsible for adding, deleting and modifying , Rarely responsible for inquiry , The database is only responsible for making queries , You are not responsible for adding, deleting, or modifying , At this time, the data in the master database will change in some way , Automatically synchronize to all slave databases , Finally, keep the data consistent

1. Master Open a thread in , Record data changes to a binary log file (binary log) in
2. Slave Open one in IO Threads are dedicated to reading Master Binary log contents in , And write the data into your own playback log file (relay log) in
3. Slave in SQL The thread reads the playback log (relay log) The content of , hold Master Implemented in DML Played back , Complete master-slave synchronization

Be careful ：

         There is a certain delay in master-slave synchronization , Therefore, there is no guarantee that the master database synchronization information can be read from the database immediately , Therefore, as long as the write operation is involved in the business method , Then the operation must be performed on the primary database

4  Implementation steps

preparation :

         First, let Master and Slave All have the same database , And the table data in the database is consistent , I use my own here skywalking library .

1、 stay Master In the configuration file [mysqld] Some of the following configurations are made , Turn on binary-log

# Configure the current server id 
server-id=1 
# Configure binary log file name  
log-bin=master-bin 
# Configure which database to log , You can configure it multiple times  
binlog-do-db=skywalking

And then restart MySQL After service , Log in again to see how to use SQL Command to view information , The following contents are displayed to show that the first step is successful

--  View the status of the primary database  
show master status; 
--  Be careful :bin-log The file name of will change with the restart of the master server 

 

2、 stay Master Create a new user in the database for master-slave synchronization

--  You can use a graphical interface to operate  
--  Create user slave For master-slave synchronization , And conferred rbac Database operation permission  
GRANT ALL ON `skywalking`.* TO 'slave'@'%' IDENTIFIED BY '1' 

--  Grant master-slave synchronization permission , This permission is a global permission  
GRANT replication slave ON *.* TO 'slave'@'%' 

--  Refresh the permissions  
flush privileges

3、 stay Slave In the configuration file [mysqld] Some of the following configurations are made , Then restart the database service

# Configure slave id, It must be the only one that cannot repeat  
server-id=2 
# Which database logs are played back , You can configure it multiple times  
replicate-do-db=skywalking

4、 Sign in Slave In the implementation of SQL

CHANGE MASTER TO 
MASTER_HOST='192.168.222.131',
MASTER_USER='slave', 
MASTER_PASSWORD='1', 
MASTER_LOG_FILE='master-bin.000001', 
MASTER_LOG_POS=154

among ：
        MASTER_HOST Primary service ip Address ,
        MASTER_USER The user name of the master-slave replication created above , 
        MASTER_PASSWORD Password for the user created above , 
        MASTER_LOG_FILE by bin-log The name of the file , 
        MASTER_LOG_POS Is the length position of the log file

Execute... On the main database

show master status; 

see bin-log File name and length of log file

File For the file name ,Position Is the length position of the log file

Be careful ：

         If the master-slave synchronization has been enabled in the slave database, you need to close it first , To execute the above configuration

stop slave

Start master-slave synchronization after configuration

start slave

Check the database status after startup ：

--  View from database status  
show slave status 

notice Slave_IO_Running and Slave_SQL_Running All for YES It means success

If there is an error , Please check the corresponding Last_Error_Log