Glosarium
Contents

Backend Development › Backend Basics · also in Relational Databases & SQL

ORM

Mapping database tables to objects in code.

Backend Engineer · Junior · core

Also known as: object-relational mapper, object relational mapping, Django ORM, SQLAlchemy, Hibernate, ActiveRecord

An ORM (object-relational mapper) lets you work with database tables as objects in your programming language, instead of writing SQL by hand. Examples: Django ORM, SQLAlchemy, Hibernate, ActiveRecord, Prisma, TypeORM.

# Django-style
class Order(models.Model):
    customer = models.ForeignKey(Customer, on_delete=models.CASCADE)
    total_cents = models.IntegerField()
    status = models.CharField(max_length=20)

big = Order.objects.filter(status="paid", total_cents__gt=10000).order_by("-id")[:10]

The ORM translates that into SQL and turns the resulting rows back into Order objects. Saving works the same way: order.status = "shipped"; order.save() runs an UPDATE.

What you gain

  • Less repetitive code for the usual create, read, update and delete.
  • Safe parameter handling, which protects against SQL injection when used properly (prepared statements).
  • Relationships as attributes (order.customer.name).
  • Portability across database engines, to a degree.
  • Often migrations, schema tools built in.

What to watch out for

The N+1 query problem. Looping over objects and touching a relationship can fire one query per row:

for order in Order.objects.all():      # 1 query
    print(order.customer.name)         # +1 query per order!

Fetch related data up front (select_related, joinedload or your ORM’s equivalent). See N+1 queries.

Not knowing the SQL. An ORM doesn’t remove the need to understand the database: indexes, transactions and joins still decide performance. Print or log the generated SQL and look at it.

Loading too much. Fetching whole objects when you need two columns, or whole tables into memory.

Leaky complexity. Complex reports and bulk operations are often easier and faster in plain SQL, and ORMs let you drop to it when needed (ORM vs raw SQL). Lighter alternatives are query builders.

Learn SQL first, or alongside. It makes you much better at using any ORM.