Chapter two of Designing Data‑Intensive Applications walks developers through how raw data becomes structured storage. After collection, data is shaped into objects or APIs and persisted as JSON/XML documents, relational tables, or graph structures. The relational model remains the workhorse, while many systems now adopt polyglot persistence to blend SQL and NoSQL stores.

SQL databases dominate developer skillsets, yet NoSQL promises higher write throughput, flexible schemas, and queries that relational engines cannot express. Object‑oriented code collides with tables, creating the infamous impedance mismatch that ORMs merely mask. Document models like MongoDB store nested JSON, delivering superior locality for one‑to‑many data and sidestepping costly joins.

Early hierarchical and network models—IBM’s IMS and pointer‑based access paths—failed to scale query complexity, paving the way for the relational model with its simple tables and powerful query optimizers. Declarative SQL lets databases parallelize work automatically. Upcoming trends include tighter integration of SQL and document stores, so developers should monitor hybrid query engines.