Amr Samir | Multi-Tenant SaaS Architecture & Isolation Patterns

1. What Is Multi-Tenancy?

A Software-as-a-Service (SaaS) application serves multiple business customers (tenants) from a shared infrastructure. The primary engineering challenge is guaranteeing strict data isolation and cost efficiency across scaling boundaries.

code

+----------------------------------------------------------------------+
|                     3 Multi-Tenancy Database Models                  |
+----------------------------------------------------------------------+
| 1. Database-per-Tenant  | Separate DB for every customer (High cost) |
| 2. Schema-per-Tenant    | Shared DB, isolated Postgres schemas       |
| 3. Shared Database      | Shared tables with TenantId column (Low cost|
+----------------------------------------------------------------------+

2. Comparing Multi-Tenancy Strategies

code

+--------------------+----------------+------------------+-------------------+
| Feature            | DB-per-Tenant  | Schema-per-Tenant| Shared Table      |
+--------------------+----------------+------------------+-------------------+
| Data Isolation     | Maximum (Hard) | High (Logical)   | Logical (Software)|
| Cost Efficiency    | Low            | Medium           | Maximum           |
| Scale Boundary     | ~ 100 tenants  | ~ 1,000 tenants  | 10,000+ tenants   |
| DB Migrations      | Very Slow      | Slow             | Fast (Single Run) |
+--------------------+----------------+------------------+-------------------+

3. Implementing Safe Shared-Table Isolation with Prisma / NestJS Middleware

typescript

import { PrismaClient } from '@prisma/client';

export function createTenantPrismaClient(tenantId: string) {
  const prisma = new PrismaClient();

  return prisma.$extends({
    query: {
      $allModels: {
        async $allOperations({ model, operation, args, query }) {
          // Automatically inject tenantId into all queries
          if (['findMany', 'findFirst', 'count'].includes(operation)) {
            args.where = { ...args.where, tenantId };
          }
          if (['create'].includes(operation)) {
            args.data = { ...args.data, tenantId };
          }
          return query(args);
        },
      },
    },
  });
}

4. Critical Security Failures in SaaS Systems

Cache Pollution: Always namespace cache keys with tenant ID (tenant:${tenantId}:user:${userId}).

Background Jobs: Always preserve and pass the tenantId in message queue payloads.

Client-Side Tenant Tampering: Never trust tenantId submitted from URL query params or request bodies without verifying against the authenticated JWT session.

Designing Multi-Tenant SaaS Architecture: 100 to 10,000 Tenants Without Data Leaks

Recommended Posts

Docker & Kubernetes — From Containerization to Cluster Orchestration

Reliable User Registration: Solving the Dual-Write Problem with Transactional Outbox

High-Scale Database Pagination: Why OFFSET Kills Performance and How Cursor Wins

Related Projects

Designing Multi-Tenant SaaS Architecture: 100 to 10,000 Tenants Without Data Leaks

1. What Is Multi-Tenancy?

2. Comparing Multi-Tenancy Strategies

3. Implementing Safe Shared-Table Isolation with Prisma / NestJS Middleware

4. Critical Security Failures in SaaS Systems

5. Summary