Developer Maneshwar explains how SQLite stores everything in a single file, treating raw bytes as structured data. The core abstraction is the page—a fixed-size chunk managed by the pager module. Every database file size must be a multiple of the page size, with pages numbered sequentially starting from page 1.

Page size itself is flexible, ranging from 512 to 65,536 bytes, and must be a power of two. Once set via PRAGMA before creating the first table, it becomes a permanent property of the file, not the library. This design allows massive theoretical databases, though filesystem limits usually intervene first.

SQLite meticulously tracks every page, classifying them as free, tree, pointer-map, or lock-byte. Tree pages form B-tree structures for data and indexes, while overflow pages handle large records. Page 1 is the anchor, housing the 100-byte file header containing the magic signature, schema info, and format version.

This byte-level precision reveals SQLite as a highly engineered storage engine. Every byte has a defined purpose, ensuring no garbage collection or orphaned space. The header dictates how the library interprets the file, making the database portable across systems regardless of the library's default settings.