MySQL master-slave replication

Data can be obtained from a MySQL The database server master node is replicated to one or more slave nodes .

The master server is only responsible for writing , The slave server is only responsible for reading

species ：

Master slave synchronization ： When the user writes data, the master server must synchronize with the slave server to tell the user that the writing is successful , The waiting time is long .

Master slave asynchrony ： As long as the user accesses the master server for writing data , Return to the user immediately .

Master slave semi synchronous ： When the user accesses and writes data, the master server writes and synchronizes one of the slave servers, and then returns success to the user .

principle ：

Data modification and writing master Database binlog in .

slave Of IO Threads copy these changes binlog To your own relay log in .

slave Of SQL The thread reads and reapplies relay log To your own database , Let it go together master Database consistency .

Log files ：

Error log (error log)： Record mysql Right and wrong information when starting and stopping the service , It also records the start of 、 stop it 、 Error messages during operation .

Query log (general log)： Record established client connections and executed statements .

Binary log (bin log)： Record all statements that change data , Can be used for Data replication .

Slow query log (slow log)： Record all execution times over long_query_time All queries of or queries without index .

relay logs (relay log)： Master slave copy Log used when .

advantage ：

Provides the ability to separate reading and writing , send MySQL The database supports high concurrent read and write

It can improve the security of data , Effectively protect the data backup of physical server downtime

Insufficient ：

Master-slave delay , Damage disaster tolerance , Because the data is out of sync

The reason for the master-slave delay

1、TPS When high , Produced DDL The quantity exceeds Slave One SQL Thread processing capability

2、Slave Medium and large Query Lock waiting occurs

How to judge whether there is a master-slave delay

show slave status -> Seconds_Behind_Master

How to avoid master-slave delay

1、 close Slave Of sync_binlog

2、 close Slave Of innodb_flushlog

3、 Performance ratio Master Better machines as Slave

4、 Set up one that does not provide query service Slave

5、 increase Slave The number of , Lower the unit Slave Reading pressure on

Interview answer ：

MySQL It is a small and medium-sized relational database , It's based on Multithreading .

There are some bottlenecks in data storage and query speed , Read write separation is one of the solutions .

Main library write , Read from library .

Hang up the main warehouse , From the library, you can go up , Avoid the loss of downtime .

The principle of data synchronization is ,

Data modification and writing master Database binlog in .

slave Of IO Threads copy these changes binlog To your own relay log in .

slave Of SQL The thread reads and reapplies relay log To your own database , Let it go together master Database consistency .

There are also shortcomings. , There are delays and increased hardware purchase costs

The forms are ,

A master from

One master, many followers

Load Balancer ,

According to the algorithm , Which one to test MySQL Low pressure , Distribute the request to this station MySQL

Be similar to nginx Proxy service for

It can only relieve certain pressure , But it can't be cured .

Better than that redis, Memory based operations , Single threaded , Fast query speed .