MySQL master-slave replication

Why copy ？ 

1. Prevent accidental loss of data , Make users suffer losses .
2. A machine is down , You can enable the backup of data on another machine .
3. The probability of downtime is very small , Spare time can also allow the backup machine to share the flow pressure of the main machine .
4. When upgrading the database version , The standby machine can be upgraded preferentially without stopping the user service , Upgrade the main database after observing its availability and stability .

Can't always let DBA Copy manually to complete the replication , So we still need to design a mechanism that can automatically copy .

Design replication mechanism

Our tentative replicated database is the main database , Pasted from the Library , To realize the replication from master library to slave Library , It looks very simple , Just one planning task , Regularly copy a copy of the main database data file , And transfer it to the server where the slave library is located .

But after all, scheduled tasks are not real-time , In case the main database fails ten minutes after the last replication , The activated slave library uses the last copied data , So ten minutes of data will be missing , The consequences are unimaginable .

Or do you want to copy in real time , That can be like this , The master library sends the executed statements to the slave Library in real time , Let the slave execute immediately , It can ensure the consistency of data on both sides .

The bad thing is , The master database sends data to the slave database in real time , You need to wait for the execution of the slave library to process the next statement , It seriously takes up the execution time of the main library , If there are too many libraries , The main warehouse is abandoned .

It has to be changed to asynchronous to save the time of the main database , You can save the executed statements of the main library into a file , Let's get it from the library , In this way, the master database does not have to wait for the slave database . Since it is written to the file , The speed is very fast , The main library can write statements to files before execution , Achieve higher synchronization efficiency .

Some of the above problems , It is impossible to run to the main database to get data from the database , Only one thread can be started to establish a connection with the main library , And ask for data from the main database , Then the main library also starts a thread to read the contents of the file , And push it to the slave thread , After receiving the statement from the library, it can be executed immediately .

Such efficiency is still very low , The thread of the main library cannot push another one until the statement is received and executed from the Library , If there are multiple slave Libraries , The master library needs to open multiple threads to keep communicating with each slave library for a long time , Occupy the main library server resources , It's better to create a file from the library to temporarily save the statements sent from the main library , Save it first and then execute it slowly , The pressure of the main reservoir is reduced , Rest assured from the Library .

Now I have my own file from the library as a relay , Don't worry , You can start another thread from the Library , Slowly execute the statements in the relay file , After the implementation, the original document has no value , You can clean it up , Avoid occupying server resources .

up to now , The most basic replication mechanism is designed , This way of copying from master database to slave database is a typical master-slave architecture , On this basis, evolution can be carried out , For example, there are many from the Library , The master database should push data for each slave database , The pressure of the main reservoir will increase , And because the main database is not only responsible for synchronizing data , Also busy reading and writing data , So people can replace the synchronization of data , For example, establish a master database between the master database and the slave database , The only responsibility of the newly established master database is to synchronize data to the slave database , In this way, the real master database only needs to push data to the new master database once , You can read and write data at ease in the rest of the time .