Why should you care?

Banks and healthcare clients audit how data flows — sloppy coupling between UI and DB queries fails reviews.

See it live — copy this example

Paste into your MeanVerse project (Angular + Express + MongoDB), then run with ng serve / node / mongosh as noted.

// api/src/modules/accounts/account.service.ts
export class AccountService {
  constructor(private repo: AccountRepository) {}
  async transfer(fromId: string, toId: string, amount: number) {
    if (amount <= 0) throw new BadRequestError('Invalid amount');
    return this.repo.transferAtomic(fromId, toId, amount);
  }
}
// route stays thin
router.post('/transfer', auth, validateTransfer, (req, res, next) =>
  accountService.transfer(req.body.from, req.body.to, req.body.amount)
    .then(r => res.json(r)).catch(next));

What happened?

• Routes parse HTTP.
• Services enforce business rules.
• Repository/Mongoose handles persistence.
• Angular never calls Mongoose directly.

Practice next

1. Draw four layers for one MeanVerse use case (transfer funds).
2. Move DB calls from route handlers into a repository class.
3. Expose DTOs to Angular — hide internal Mongo _id shapes if needed.
4. Add an interface IAccountRepository for mock tests.
5. Map Mongoose docs to plain AccountDto before res.json.

Remember

Thin routes, fat domain services, isolated data access. Angular talks HTTP, never Mongo wire protocol. Boundaries simplify testing and security reviews.

Banking transfer audit

Regulators ask where transfer limits are enforced.

Outcome: AccountService is the single answer — not scattered in UI.