Designing for Multiple Customers

In SaaS, you build one system that serves thousands of customers. This is called Multitenancy. The challenge is ensuring that Customer A never sees Customer B's data.

1. Multitenancy Models

• Database-per-Tenant: Highest isolation. Every customer has their own physical DB. expensive but extremely secure and easy to back up individually.
• Schema-per-Tenant: One DB server, multiple schemas. Good balance of cost and isolation.
• Shared Database (Row-Level): Every table has a `TenantId` column. The most cost-effective but requires rigorous coding (Global Query Filters in EF Core) to prevent data leaks.

2. The "Noisy Neighbor" Problem

If Customer A runs a giant report that consumes all the CPU, Customer B's experience will suffer. To scale, you must implement **Resource Quotas** and **Rate Limiting** at the tenant level to ensure one customer's heavy usage doesn't crash the party for everyone else.

4. Career Mastery

Q: "Which and why?"

Architect Answer: "Start with **Shared Database (Row-Level)** for your MVP. It is the easiest to manage and the cheapest to run. Use **Entity Framework Core Global Query Filters** to automatically inject `WHERE TenantId = X` into every query. Only move to 'Database-per-Tenant' when an Enterprise customer with a $100k contract demands it for compliance reasons."