Relational vs NoSQL Databases: Data Models, Consistency and Scaling

Explain how relational (SQL) databases differ from NoSQL databases. ```hint Go past the label "Relational versus non-relational" is only the starting point. Compare how each one models data, where its schema is enforced, what it guarantees when many users write at once, and how it grows with traffic. ``` ### Clarifying Questions - Should I treat NoSQL as one group, or walk through its main kinds: key-value, document, wide-column and graph stores? - Do you want the engineering mechanics, or how the choice affects a product's features, data quality and delivery speed? - Would it help to ground the comparison in one product, such as an online store's checkout and catalog? ### What a Strong Answer Covers - Data model and schema: tables and joins versus the NoSQL models, and a schema enforced by the database versus by application code - Transaction and consistency guarantees, and what weaker guarantees look like to a user - How each kind of database scales, without the claim that relational databases cannot scale out - Query flexibility versus designing around known access patterns - A concrete example of when each fits, and why many products use both ### Follow-up Questions - An engineer proposes moving the payments ledger to a document store to ship faster. What would you ask? - Which features of a social app can tolerate eventual consistency, and which cannot? - When is adding a second database to a product worth the extra operational cost?