Databases as of Oct 2026

Database Tier List

Rank 62 databases from 47 makers: engines like PostgreSQL, MySQL, SQLite, MongoDB, and Redis, hosted services from Neon and PlanetScale to RDS and DynamoDB, plus search engines, vector databases, and cloud warehouses. Pick what you are building under Best for, open any spec sheet, and trace each database to the engine underneath, its data model, and its license history below.

0 of 37 ranked
Drop databases here
Drop databases here
Drop databases here
Drop databases here
Drop databases here
Drop databases here

Drag databases into tiers, or tap one and then a tier. Keyboard: focus a database, press 1–6 to rank it and 0 to unrank, or Enter then the arrow keys to move it. Ctrl+Z undoes.

Unranked

Data model
Who runs it
Engine family
License
Free tier
Features
Regions in
Origin
Company
Best for
Sort

Popular is every database in the DB-Engines top 40 (Oct 2026), open-source engines with more than 30,000 GitHub stars (Oct 2026), and the hosted databases developers start on most: Neon, Amazon RDS and Aurora, Cloud SQL, PlanetScale, Firestore, Upstash, and Turso. Switch to Engines, Hosted, or All for 25 more.

What engine is it, really?

Hosted services rarely invent a new engine: most run Postgres, MySQL, SQLite, or Redis, or speak their protocol on top of something else.

13 of the 62 databases are Postgres, run it, or speak its protocol, more than twice as many as any other single engine. 17 are NoSQL stores spread across many engines, and 6 belong to the MySQL family.

The Postgres family

PostgreSQL

1996–today

PostgreSQL Global Development Group. Still developed

Born as POSTGRES at UC Berkeley in 1986. Its extensions and wire protocol made it the default that new databases copy: most hosted services here run it or speak it.

The MySQL family

MySQL

1995–today

Oracle. Still developed

The M in LAMP. Sun bought it in 2008 and Oracle inherited it in 2010, which is why MariaDB exists. Vitess, PlanetScale, TiDB, and Aurora speak its protocol.

MariaDB

2009–today

MariaDB plc & MariaDB Foundation. Still developed

Monty Widenius forked MySQL when Oracle was about to own it. Still a drop-in replacement for most apps, and the default MySQL in many Linux distros.

In-process & embedded

SQLite

2000–today

D. Richard Hipp. Still developed

A database in one file, linked into the app. The most deployed database engine there is, now also served at the edge by Cloudflare D1.

libSQL & Turso

2022–today

Turso. Still developed

An open-contribution fork of SQLite with replication, now followed by Turso Database, a from-scratch Rust rewrite that stays file-compatible with SQLite.

DuckDB

2019–today

DuckDB Labs & CWI Amsterdam. Still developed

An in-process columnar engine for analytics: SQLite’s idea applied to OLAP, querying Parquet, CSV, and Postgres directly.

Commercial & distributed SQL

SQL Server & T-SQL

1989–today

Microsoft. Still developed

Started from a Sybase codebase in 1989. Azure SQL Database runs the same engine as a service.

Oracle Database

1979–today

Oracle. Still developed

The first commercial SQL database, still at the top of the DB-Engines ranking.

Spanner & distributed SQL

2012–today

Google. Still developed

Google’s globally consistent SQL database. Its 2012 paper inspired the distributed SQL wave: CockroachDB, YugabyteDB, and TiDB.

Key-value & cache

Redis

2009–today

Redis Ltd. Still developed

The in-memory data structure server. Its protocol is spoken by Valkey, Dragonfly, Upstash, and every cache service.

Valkey

2024–today

Linux Foundation. Still developed

The BSD-licensed fork of Redis 7.2, started by former Redis maintainers with AWS, Google, and Oracle when Redis left open source in 2024.

Documents

MongoDB & its wire protocol

2009–today

MongoDB Inc.. Still developed

The document database that defined the NoSQL era. FerretDB and several cloud services speak its protocol on top of other engines.

CouchDB & Couchbase

2005–today

Apache & Couchbase. Still developed

CouchDB’s JSON-over-HTTP and sync model, and Couchbase, which grew from a 2011 merger of Membase and CouchOne.

Multi-model newcomers

2012–today

Various. Still developed

Databases that mix documents, relations, and graphs behind their own query languages.

Wide-column & cloud NoSQL

Cassandra & CQL

2008–today

Apache Software Foundation. Still developed

Facebook’s wide-column store, built from Amazon’s Dynamo and Google’s Bigtable designs. ScyllaDB rewrote it in C++.

Hyperscaler NoSQL

2008–today

AWS, Google & Microsoft. Still developed

Proprietary NoSQL services that only run on their own cloud: DynamoDB, Firestore, Bigtable, and Cosmos DB.

Search

Elasticsearch

2010–today

Elastic. Still developed

Distributed search and analytics on Apache Lucene, the E in the ELK stack.

OpenSearch

2021–today

OpenSearch Software Foundation. Still developed

AWS forked Elasticsearch 7.10 when Elastic dropped the Apache license, and moved it to the Linux Foundation in 2024.

Developer search engines

2012–today

Various. Still developed

Typo-tolerant, instant search built for app search boxes rather than log analytics.

Vectors, graphs & time

Vector databases

2019–today

Various. Still developed

Built for nearest-neighbor search over embeddings, the retrieval half of RAG. Postgres (pgvector) and most engines here added vectors too.

Graph databases

2007–today

Various. Still developed

Nodes and relationships queried with Cypher, now standardized as ISO GQL.

Time-series databases

2013–today

Various. Still developed

Optimized for timestamped metrics and events: fast ingest, downsampling, and time-window queries.

Analytics

Columnar analytics & warehouses

2011–today

Various. Still developed

Column stores that scan billions of rows: cloud warehouses that separate storage from compute, and ClickHouse.

Dashed logos mark databases that also run, speak, or descend from that engine.