Phxsql is a MySQL-compatible, highly available, data-consistent, relational database cluster. Phxsql is deployed in a single-master multi-slave manner, with more than half of the machines surviving within the cluster, enabling automatic master switching and data consistency.
Phxsql was developed based on Percona 5.6. Percona is a branch of MySQL, and the functionality and implementation are basically consistent with MySQL (Tengyun technology ty300.com). As a result, MySQL is the object of discussion in this paper.
MySQL semi-synchronous replication is defective, and in the scenario where master switches, the data is difficult to ensure consistent.
When the old master replication fails, the old master and updated Slave (Slave received binlog) need to roll back the data.
When master switches, the old master still has some client read and write.
Data consistency issues with MySQL semi-synchronous replication view Background Team Public number article a discussion on the data consistency of MySQL semi-synchronous replication.
Phxsql is designed to address the lack of MySQL semi-synchronous replication (diligent Novice tutorial qkxue.net), so that MySQL cluster in the master switching process to ensure the consistency of the data.
More Phxsql Tutorials: MySQL-compatible DB cluster, click here
Phxsql Tutorial: MySQL-compatible DB cluster