Overview
PostgreSQL is the world's most-loved open-source relational database, governed by the PostgreSQL Global Development Group and licensed under the permissive PostgreSQL Licence. Version 17 (released September 2024) brought incremental backup, an improved VACUUM that uses 20x less memory, native MERGE improvements, and a more efficient streaming I/O subsystem. Version 18 entered beta in early 2026 and is expected to ship the long-awaited asynchronous I/O implementation.
For commercial buyers, the relevant choices are: self-manage the community release, contract support from EDB (the largest contributor outside the core team), Crunchy Data, or Percona, or consume PostgreSQL as a managed cloud service — Amazon RDS for PostgreSQL, Amazon Aurora PostgreSQL-Compatible, Google Cloud SQL, Azure Database for PostgreSQL — Flexible Server, or specialty vendors like Neon, Supabase, Tembo, and Crunchy Bridge. PostgreSQL has been the default new-application database for most enterprises since around 2020 and remains the fastest-growing database in the Stack Overflow Developer Survey.
Key Features
- ACID-compliant relational engine with sophisticated MVCC concurrency control
- Native JSON and JSONB document storage with GIN indexing
- Logical and streaming physical replication; native publications and subscriptions
- Extensive procedural language support — PL/pgSQL, PL/Python, PL/Perl, PL/v8
- Foreign Data Wrappers for federated queries against Oracle, MySQL, MongoDB, S3, files
- Rich extension ecosystem — PostGIS, pgvector, TimescaleDB, Citus, pg_partman
- Partitioning (range, list, hash) with partition pruning and parallel query
- Row-level security, column encryption, SCRAM-SHA-256 authentication
- Generated columns, identity columns, window functions, CTEs, recursive queries
- Logical decoding for change data capture pipelines (Debezium, Kafka Connect)
- Incremental backup with pg_basebackup --incremental (Postgres 17)
- pgvector extension provides production-grade vector similarity search
Pricing
| Option | Model | Indicative Cost |
|---|---|---|
| PostgreSQL community release | Open source (PostgreSQL Licence) | $0 |
| EDB Standard support | Per instance/year | From ~$1,300/instance/year |
| EDB Enterprise (EDB Postgres Advanced Server) | Per core/year | ~$2,000–4,000/core/year |
| Crunchy Data Postgres support | Per cluster/year | From ~$25K/cluster/year |
| Amazon RDS for PostgreSQL | Per instance/hour | From ~$0.018/hour (db.t4g.micro) to $10+/hour |
| Azure Database for PostgreSQL Flexible Server | Per vCore/hour | From ~$0.034/vCore/hour (Burstable) |
Pricing verified May 2026. EDB and Crunchy pricing requires direct quote and is highly negotiable based on contract size. Managed cloud pricing assumes on-demand; reserved or savings-plan commitments reduce cost 30–60%.
Strengths
- Permissive licence with no commercial restrictions and broad community governance
- Strongest extension ecosystem of any relational database (PostGIS, pgvector, TimescaleDB)
- SQL standards compliance is among the highest in commercial or open-source databases
- Mature replication, partitioning, and HA tooling (Patroni, repmgr, pgpool-II)
- Available as a first-class managed service on every major cloud platform
- Strong reputation for stability, durability, and conservative release engineering
Limitations
- MVCC creates table bloat — VACUUM tuning is a recurring operational task
- Built-in HA is limited; production HA requires third-party tools (Patroni, repmgr)
- Logical replication has known gaps around DDL, sequences, and large transactions
- No built-in multi-master replication — read-write scale-out requires Citus or BDR
- Connection scaling is weak without a pooler (PgBouncer, pgcat) in front