SQL & Database2 min readUpdated August 7, 2026
Types of Database: PostgreSQL, MySQL, SQL Server and the Rest
A map of the databases in common use and what each is like — plus one term that is confused more than any other: SQL is a language, not the name of a database.
PostgreSQLMySQLSQL ServerOracleSQLiteMongoDBRedisNoSQL
The big division is relational versus non-relational. Relational databases store data in tables of rows and columns with a schema defined up front, and maintain the relationships between tables. Non-relational databases (often called NoSQL) relax those rules in exchange for a freer data shape or higher speed.
The relational ones
| Database | Character | Commonly used for |
|---|---|---|
| PostgreSQL | Open source, standards-faithful, strong on complex queries and data types such as JSON | Applications with complex data that has to be correct |
| MySQL / MariaDB | Open source, light, the widest hosting ecosystem | General web applications, CMSs, Laravel projects |
| Microsoft SQL Server | Commercial, tightly integrated with the Microsoft ecosystem | Organisations already on a Microsoft stack |
| Oracle Database | Commercial, enterprise grade, strong at large volumes | Banking and large-scale systems |
| SQLite | A single file, no separate server | Mobile and desktop applications, and testing |
The non-relational ones
| Database | Data shape | Commonly used for |
|---|---|---|
| MongoDB | JSON-like documents | Data whose structure keeps changing |
| Redis | In-memory key-value pairs | Caching, sessions, queues |
| Elasticsearch | Documents indexed for search | Text search and log analysis |
| Cassandra | Wide-column, distributed | Very large data with heavy write volume |
How to choose
- The data has clear relationships between entities and involves money — choose relational. Transactional guarantees are worth more than schema flexibility.
- The structure is not settled and every document may hold different fields — a document store such as MongoDB makes sense.
- What you need is extreme read speed for transient data — Redis, while keeping the source of truth in the primary database.
- You are already committed to an ecosystem — follow what is there. Adding a database type adds something to maintain, back up and restore.
For the majority of business applications, PostgreSQL or MySQL is the safe answer. Both can now store JSON as well, so the old argument that flexible schemas require NoSQL no longer holds automatically.