CSE Roadmap

Common tier · about 2 hours · 5 parts

Part 2 of 5 · 1 min read

Row-oriented storage

All of one record's values stored together, physically adjacent on disk.

Think of an index card system for customers. Each card holds everything about one customer: name, address, every detail, all on one card.

In plain words: row-oriented storage keeps all of one row's column values physically adjacent on disk. Reading one whole row, such as for a single order's full details, is efficient, since everything needed sits in one place. Reading just one column across many rows, such as "every order's placed_on date," requires touching every row's full data anyway.

A worked example. Fetching one specific order's complete details, for a receipt, reads efficiently from row-oriented storage, since the whole row is contiguous.

The key lines.

SELECT * FROM orders WHERE order_id = 9001;
-- efficient in row-oriented storage: the whole row is stored together

Row-oriented storage

Row-oriented storage keeps each row's full set of column values physically adjacent, efficient for reading or writing whole records at a time.

Example: fetching one order's complete row for a receipt is a natural fit for row-oriented storage.