ORMs

An object-relational mapper (ORM) is a software library that maps between in-memory objects and relational database tables. It provides an object-oriented interface for data access while translating operations to SQL under the hood, reducing boilerplate queries and manual result handling.

Core concepts include models or entities that correspond to database tables, attributes that map to columns, and relationships that represent foreign keys (one-to-one, one-to-many, many-to-many). Most ORMs implement patterns such as the identity map and unit of work to track changes to objects and persist them within a single transaction. They typically manage database connections, generate SQL, offer a query API, and may provide schema migrations to evolve the database alongside code.

How it works: developers define classes and mappings; the ORM builds SQL, executes it, and materializes results into objects. It can support lazy loading, eager loading, and caching. Migrations help keep the database schema in sync with the application's data models.

Advantages include faster development, cleaner code, and improved portability across databases. Disadvantages include potential performance pitfalls (notably N+1 queries), abstraction leakage that hides SQL behavior, reduced control over specialized queries, and complexity with intricate data access patterns.

Common examples are Hibernate (Java), SQLAlchemy (Python), Django ORM (Python), Entity Framework (C#), ActiveRecord (Ruby on Rails), and Eloquent (PHP). ORMs are well-suited for CRUD-heavy applications and rapid development when the data model aligns with object-oriented design; micro-ORMs offer lighter-weight alternatives when finer control is needed.