What is the N+1 query problem and how do you fix it?

The first query returns N rows. A lazy relationship then triggers a query for each row, producing N+1 round trips. The result may be correct, but latency grows linearly and the database sees far more queries than necessary.

const users = await db.user.findMany();

for (const user of users) {
  user.orders = await db.order.findMany({
    where: { userId: user.id },
  });
}

The fixes

• Use a join or an eager-loading option that fetches the relation in one query
• Batch the child ids into a single IN query and group the results in memory
• Use a dataloader in GraphQL or similar APIs to coalesce requests per tick
• Select only the columns you need to reduce payload and memory
• Add an index on the foreign key used by the child lookup
• Watch nested serializers, which can reintroduce N+1 outside the original query

Eager loading is not automatically better. Joining a one-to-many relation can duplicate the parent row, so some ORMs split it into two queries. That is still fine if it is a constant number of queries. The goal is bounded query count, not a single query at any cost.