Why Choose PostgreSQL Over MySQL Despite MySQL’s Popularity?

Domestic PostgreSQL‑based databases

Tencent TDSQL PG (open source name: TBase) – https://github.com/Tencent/TBase Alibaba PolarDB for PostgreSQL – distributed, cloud‑native storage with "one‑write‑many‑read" shared storage.

Huawei GaussDB (compatible with the openGauss ecosystem) – columnar engine and AI optimizer for HTAP workloads – https://opengauss.org Hangzhou Yijing openHalo – open‑source HTAP platform –

https://github.com/HaloTech-Co-Ltd/openHalo

Why PostgreSQL is preferred over MySQL

Data‑type richness

Array types – ARRAY stores multiple values in a single column.

Range types – e.g., int4range, tsrange represent numeric or temporal intervals.

Composite types – custom structures such as POINT(x,y) model real‑world objects.

JSONB – binary JSON with indexing, query and update capabilities, delivering far better performance than plain JSON fields.

Native sequence support

CREATE SEQUENCE order_seq START WITH 1 INCREMENT BY 1;

INSERT INTO orders (id, name) VALUES (nextval('order_seq'), 'test');

MySQL lacks independent sequence objects. It simulates sequences with AUTO_INCREMENT plus a large offset, and the sequence is bound to a specific table, making cross‑table sharing impossible without external middleware.

ALTER TABLE orders AUTO_INCREMENT = 1000;

Extension ecosystem

TimescaleDB

– time‑series database with automatic partitioning and compression. pg_trgm – fuzzy matching and similarity search. Citus – distributed database extension. pg_stat_statements – SQL execution statistics and monitoring.

Monitoring and diagnostics

Built‑in statistical views: pg_stat_activity, pg_stat_statements, pg_locks. EXPLAIN ANALYZE shows actual execution time and row counts.

Third‑party tools with mature integrations: PgAdmin, pg_stat_monitor, Prometheus + Grafana.

Replication and high availability

MySQL default asynchronous replication can cause lag; GTID solves part of the problem but is error‑prone; semi‑sync replication requires extra configuration and has high network demands.

PostgreSQL provides streaming replication (asynchronous or synchronous), logical replication per table or across versions, a mature WAL mechanism for durability, and synchronous replication that waits for at least one standby before committing, achieving zero data loss.

License and community

MySQL – GPL + commercial license controlled by Oracle; enterprise edition adds audit, encryption, etc.

PostgreSQL – BSD‑like license, fully free; community edition provides the complete feature set.

MySQL development is steered by Oracle, limiting community contributions; PostgreSQL is maintained by a global developer community with transparent governance.

MySQL roadmap can shift with corporate decisions (e.g., Oracle’s acquisition of Sun); PostgreSQL is governed by a foundation independent of any single company.

MVCC implementation

PostgreSQL stores multiple versions per row in the heap; old versions remain until vacuum cleans them, providing full read/write isolation and serializable snapshot isolation.

MySQL keeps only the current version in the row; older versions reside in the undo log, offering faster reads but risking undo‑log bloat for long transactions.

Result: In PostgreSQL, readers can see a previous state while a writer updates; in MySQL, uncommitted changes may appear as dirty reads or non‑repeatable reads depending on the isolation level.

Comparative dimensions

License : MySQL – GPL + commercial (Oracle); PostgreSQL – BSD‑like, free.

Enterprise vs community : MySQL enterprise adds features such as audit and encryption; PostgreSQL community edition includes the full feature set without cuts.

Source transparency : MySQL core development is Oracle‑controlled; PostgreSQL is maintained by a worldwide community with open source transparency.

Long‑term stability : MySQL roadmap may change with corporate decisions; PostgreSQL is led by a foundation, ensuring independence from any single vendor.

Feature timeline note

MySQL 8.0 introduced window functions only in version 8.0, whereas PostgreSQL had them much earlier.