MySQL vs PostgreSQL

Quick verdict

MySQL is developed by Oracle. The MySQL Community Server is free under the GPL v2, and Oracle also sells commercial editions (Enterprise, Standard and NDB Cluster CGE among them) and runs the MySQL HeatWave cloud service. MySQL has two release tracks: LTS releases (8.4 and 9.7 are the current LTS lines) and Innovation releases. In July 2026 MySQL moved to calendar versioning, and MySQL 26.7.0 is the first such release; it is an Innovation release.

PostgreSQL is developed by the PostgreSQL Global Development Group and released under the PostgreSQL Licence, a permissive licence similar to BSD and MIT. A new major version ships roughly once a year and each is supported for five years. PostgreSQL 18 is the current major version (18.6 at the time of writing), and versions 14 to 18 are supported, with 14 reaching end of life in November 2026.

Side by side

Key differences

Licensing and who controls the project

Oracle makes the MySQL server and client libraries available under the GPL v2 (with an additional permission for included software under other licences) and sells commercial licences for software that embeds or bundles MySQL without GPL obligations. Oracle also offers commercial editions such as MySQL Enterprise Edition. PostgreSQL uses the PostgreSQL Licence, which permits use, copying, modification and distribution without fees or written agreements, and the project states it has no plans to change it.

For most teams running a database behind their own application, both licences are free to use. The difference matters mainly if you distribute software that includes the database. This page describes licence terms; it is not legal advice.

JSON vs JSONB

MySQL's JSON type validates documents on insert and stores them in an internal binary format for quick access to elements. JSON columns are not indexed directly: you index a generated column or use a functional index, and InnoDB supports multi-valued indexes on JSON arrays. MySQL can also apply partial in-place updates through JSON_SET(), JSON_REPLACE() and JSON_REMOVE() under documented conditions.

PostgreSQL has two types. json stores an exact copy of the input text; jsonb stores a decomposed binary form that drops whitespace and key order, keeps only the last of duplicate keys, and supports GIN indexes. The PostgreSQL documentation recommends jsonb for most applications.

MVCC and locking

Both engines use multi-version concurrency control so that readers see a consistent snapshot. InnoDB keeps hidden transaction and roll-pointer fields on each row and rebuilds older versions from undo logs; deleted rows are physically removed later by the purge process. Its default isolation level is REPEATABLE READ, where range searches take gap or next-key locks to block inserts into the scanned range.

In PostgreSQL an UPDATE or DELETE leaves the old row version in the table until VACUUM reclaims it, and the documentation states that reading never blocks writing and writing never blocks reading. The default isolation level is READ COMMITTED, and SERIALIZABLE is implemented with Serializable Snapshot Isolation. Autovacuum is enabled by default; the docs advise against disabling it.

Replication and high availability

MySQL replication is asynchronous by default, with semisynchronous and delayed replication, and GTID-based replication that avoids tracking log file positions. Group Replication adds single-primary mode with automatic primary election, or multi-primary mode. The manual notes that Group Replication has no built-in client failover and recommends InnoDB Cluster, which combines it with MySQL Router and MySQL Shell.

PostgreSQL provides WAL-based streaming replication (optionally synchronous), hot standby for read-only queries on replicas, and per-table logical replication. Its documentation is explicit that PostgreSQL does not provide the software to detect a primary failure and promote a standby automatically; you trigger promotion with pg_ctl promote or pg_promote(), or use a third-party HA tool. Managed cloud services handle this for you on both engines.

Extensions: PostGIS, pgvector and MySQL's built-in types

PostgreSQL's extension system is one of its main draws. PostGIS adds storage, indexing and querying of geospatial data. pgvector adds vector similarity search with HNSW and IVFFlat indexes and supports PostgreSQL 13 and later. Managed services such as Azure Database for PostgreSQL document pgvector support.

MySQL builds spatial types into the server (InnoDB and MyISAM support SPATIAL indexes) and has a VECTOR data type holding up to 16,383 single-precision values. A VECTOR column cannot be used as a key. Check the MySQL documentation for which vector search functions are available in your edition before planning on them.

SQL dialect: side by side

Auto-increment vs identity. MySQL uses AUTO_INCREMENT; PostgreSQL supports the SQL standard identity column.

-- MySQL
CREATE TABLE customers (
  id    INT AUTO_INCREMENT PRIMARY KEY,
  email VARCHAR(255) NOT NULL UNIQUE,
  name  VARCHAR(100)
);

-- PostgreSQL
CREATE TABLE customers (
  id    integer GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
  email varchar(255) NOT NULL UNIQUE,
  name  varchar(100)
);

Upsert. MySQL updates the row when the insert would violate any PRIMARY KEY or UNIQUE index; the current manual recommends a row alias over the deprecated VALUES() function. PostgreSQL names the conflict target, which needs a unique index or constraint, and refers to the proposed row as EXCLUDED. PostgreSQL also supports MERGE.

-- MySQL
INSERT INTO customers (email, name)
VALUES ('ana@example.com', 'Ana') AS new
ON DUPLICATE KEY UPDATE name = new.name;

-- PostgreSQL
INSERT INTO customers (email, name)
VALUES ('ana@example.com', 'Ana')
ON CONFLICT (email) DO UPDATE SET name = EXCLUDED.name;

String aggregation. MySQL's GROUP_CONCAT output is truncated at group_concat_max_len, which defaults to 1024; PostgreSQL's string_agg takes the delimiter as an argument and an optional ORDER BY inside the call.

-- MySQL
SELECT customer_id,
       GROUP_CONCAT(product ORDER BY product SEPARATOR ', ') AS products
FROM orders
GROUP BY customer_id;

-- PostgreSQL
SELECT customer_id,
       string_agg(product, ', ' ORDER BY product) AS products
FROM orders
GROUP BY customer_id;

Limiting rows. Both accept LIMIT ... OFFSET. MySQL also accepts LIMIT offset, count; PostgreSQL also accepts the SQL standard FETCH FIRST form.

-- MySQL and PostgreSQL
SELECT id, name FROM customers ORDER BY id LIMIT 10 OFFSET 20;

-- MySQL only
SELECT id, name FROM customers ORDER BY id LIMIT 20, 10;

-- PostgreSQL (SQL standard form)
SELECT id, name FROM customers ORDER BY id
OFFSET 20 ROWS FETCH FIRST 10 ROWS ONLY;

For function-by-function differences, see the SQL functions library.

Pricing and licensing

MySQL: MySQL Community Server is free under the GPL v2. Oracle sells commercial licences for embedding MySQL in products distributed without GPL obligations, and commercial editions (MySQL Enterprise Edition, Standard Edition and NDB Cluster CGE are listed on mysql.com). We did not record prices for these editions; check Oracle's MySQL store or ask Oracle sales for current terms. MySQL HeatWave is Oracle's managed cloud service, priced separately.

PostgreSQL: free under the PostgreSQL Licence for any use, with no commercial edition from the core project. Commercial support and distributions are available from third-party companies.

Managed cloud: both engines are offered by Amazon RDS (and Aurora with MySQL or PostgreSQL compatibility), Microsoft Azure and Google Cloud SQL. Cloud pricing depends on region, instance size, storage and HA options; use each provider's pricing calculator.

Pricing checked on the vendors' official pages on 7 October 2026. Prices change; confirm before buying.

Where each one leads

Limitations

When to choose each

When neither is right

• You need an embedded, serverless database inside an app or device: SQLite is usually the better fit. See SQLite vs MySQL.
• Your organisation is standardised on Microsoft tooling and licensing: consider SQL Server. See MySQL vs SQL Server and SQL Server vs PostgreSQL.
• Your data is mostly schemaless documents with no relational joins: a document database may suit better. See MongoDB vs PostgreSQL.
• You want MySQL compatibility with a community-governed fork: look at MariaDB. See MariaDB vs MySQL.

Final recommendation

Frequently asked questions

Sources

• MySQL Reference Manual: release model and calendar versioning
• MySQL licensing (GPL v2 and commercial licences)
• MySQL Reference Manual (InnoDB, JSON, replication, Group Replication)
• PostgreSQL versioning policy
• PostgreSQL Licence
• PostgreSQL documentation (MVCC, JSON types, replication, INSERT)
• PostGIS
• pgvector (GitHub)
• Amazon RDS
• Google Cloud SQL documentation

Checked October 2026.

How we research comparisons: our editorial method.