MySQL High‑Availability Secrets: Real DBA Pitfalls and Practical Solutions

This article is part of the 3306π interview series featuring Zhang Chong, DBA manager at Mafengwo and speaker at the 2019 Beijing "MySQL Fun Talk" session.

Interview Highlights

Sharing the "pits" encountered while implementing MySQL high‑availability solutions (PXC and MGR) at Mafengwo.

Introducing the design of Mafengwo's middleware – DB Man.

Discussing the "grounded" architecture of related database tools.

Zhang explains his transition from SQL Server to MySQL DBA, emphasizing the need for DBAs to master multiple databases in modern companies. He advises focusing on relational databases for deeper understanding, prioritizing learning based on company architecture needs, and considering systematic training for efficient skill acquisition.

When asked about the similarities between MySQL and SQL Server, Zhang notes that both support logical replication, but MySQL’s implementation offers higher efficiency while SQL Server provides greater flexibility. Understanding these differences helps choose the right technology for specific scenarios.

Regarding Mafengwo’s use of MySQL, he states that MySQL is the core database supporting various services, with high‑availability achieved through PXC and MGR clusters (over a hundred instances). He highlights the importance of monitoring replication group members via performance_schema.replication_group_members and warns about potential issues when node status mismatches occur.

Overall, Zhang stresses that soft skills such as architectural thinking and experience often outweigh pure technical expertise, and he shares several practical insights to help DBAs avoid common pitfalls.