Why a Former MySQL Engineer Says PostgreSQL Is the Better Open‑Source RDBMS

Background

Steinar Gunderson, former Oracle chief software engineer and member of the MySQL optimizer team, announced his departure in a blog post (2021‑12‑05). In the post he reflected on five years of work on MySQL and recommended that developers evaluate PostgreSQL as an alternative.

Critique of MySQL Architecture

Optimizer design : Described as primitive and tightly coupled with the executor, lacking modular components that would allow independent improvement.

Executor implementation : A monolithic code base where many components are interwoven, making it difficult to isolate or replace functionality.

Missing features : Full outer joins are not supported; the “slice” system (partition‑wise execution) cannot be used, limiting complex query patterns.

Refactoring constraints : Although MySQL 8.0 introduced many improvements over 5.7, Gunderson states that large‑scale refactoring resources are no longer available within Oracle.

MariaDB Fork and 2016 Redesign

MariaDB originated from a fork after Michael Widenius left Oracle, inheriting many of MySQL’s architectural shortcomings.

In 2016 MariaDB performed a complete rewrite that replaced the original MySQL code base.

The rewrite added new storage engines, extended SQL dialects, enhanced monitoring, and introduced a machine‑learning‑based workload analysis component.

PostgreSQL as an Alternative

Modular optimizer and executor architecture that more closely follows modern database design principles.

Native support for full outer joins, advanced geospatial queries, and full‑text search.

Configuration is largely static (single postgresql.conf file) and does not require extensive runtime tuning.

Migration tools exist for moving from Oracle or MySQL to PostgreSQL (e.g., open‑source migration scripts referenced by industry practitioners).

Real‑world adoption examples: Uber migrated from MySQL to PostgreSQL and later back; Square adopted PostgreSQL for geospatial and search workloads; many engineers cite lower operational overhead and stronger community support.

Considerations for Database Choice

Selection often depends on existing engineering expertise, migration cost, and specific feature requirements rather than a single “best” system. Large‑scale migrations are justified only when the target database offers clear technical advantages that outweigh the effort.

References

Steinar Gunderson’s blog: https://blog.sesse.net/blog/tech/2021-12-05-16-41_leaving_mysql.html

The Register article: https://www.theregister.com/2021/12/06/mysql_a_pretty_poor_database/