What is a relational database and how is it different from other database types?

A relational database stores structured data in related tables and uses keys to connect those tables safely. That makes joins, normalization, and integrity enforcement practical. Other database types are optimized for different shapes, such as documents, simple key-value access, or graph traversal.

Anchor the answer around tables, keys, relationships, and the workloads where integrity matters.

Explain ACID and why it matters in production systems.

ACID stands for Atomicity, Consistency, Isolation, and Durability. In practice, it means critical transactions either succeed fully or fail safely, keep the database valid, isolate conflicting work, and survive crashes once committed. This matters wherever partial writes would break the business.

Give one concrete example, such as a bank transfer or inventory reservation.

Why do teams normalize relational schemas?

Normalization reduces redundancy by storing each fact in the correct table and connecting tables through keys. That improves consistency, keeps writes cleaner, and makes integrity easier to enforce. Most systems start normalized and only denormalize when real workload pressure justifies it.

Use the customer-and-orders example because it is easy to explain quickly.

When would you choose a non-relational database instead?

Choose a non-relational system when the workload is dominated by document-heavy payloads, simple key-value scale, or graph-style traversal. The decision should come from the data model and access pattern, not fashion.

Show that you understand tradeoffs rather than taking a dogmatic position.