MongoDB is fantastic for hacking together a prototype. You don't need a schema, you just dump JSON and move on.
But as our SaaS platforms scaled, "Schemaless" became a nightmare.
The Integrity Problem
In a SaaS app, data is highly relational.
- Users have Teams.
- Teams have Projects.
- Projects have Invoices.
Handling these relationships in NoSQL requires complex application-level logic. If a developer makes a mistake, you get Orphaned Data (e.g., invoices that belong to a deleted team).
Enter PostgreSQL + Drizzle
PostgreSQL enforces these relationships at the database level using Foreign Keys.
If we try to delete a Team, the database stops us until we decide what to do with their Projects (CASCADE delete or restrict).
-- SQL enforces safety automatically
ALTER TABLE projects
ADD CONSTRAINT fk_team
FOREIGN KEY (team_id) REFERENCES teams (id)
ON DELETE CASCADE;
The Developer Experience
With TypeScript and SQL, our data structure is strictly defined. We know exactly what an object looks like. No more user?.address?.city checks everywhere.
Stability is worth the extra setup time.