How to separate business logic from infrastructure code

Business logic is the code that encodes what your application actually does — the rules, calculations, and decisions. Infrastructure is everything the application depends on to run: databases, HTTP clients, email services, file systems.

The rule: business logic should never directly import infrastructure code. This one rule makes business logic independently testable without a real database or network.

// Bad — business logic directly depends on infrastructure
function applyLoyaltyDiscount(userId) {
  const user = db.query('SELECT * FROM users WHERE id = ?', userId);
  if (user.orders_count > 10) {
    db.query('UPDATE users SET discount = 0.15 WHERE id = ?', userId);
  }
}

Separate the logic from the data access:

// Pure business logic — no infrastructure
function calculateLoyaltyDiscount(ordersCount) {
  return ordersCount > 10 ? 0.15 : 0;
}

// Infrastructure layer calls the logic
async function applyLoyaltyDiscount(userId) {
  const user = await userRepository.findById(userId);
  const discount = calculateLoyaltyDiscount(user.ordersCount);
  await userRepository.updateDiscount(userId, discount);
}

Now calculateLoyaltyDiscount is a pure function you can test with zero setup. The infrastructure layer handles all I/O and calls the pure logic with real data.

This pattern — often called hexagonal or ports-and-adapters architecture — scales from small functions to entire applications. You don't need a framework to apply it: just stop letting business logic directly call databases and HTTP clients.