From Battleship Valves to Billions of Devices

The Battleship and the Bytecode

D. Richard Hipp created SQLite in 2000 to solve a reliability problem on the USS Oscar Austin. The battleship’s software relied on Informix, but the server would fail, and the crew could not connect to the database when the ship took damage. Because the battleship software required high reliability, Hipp developed a byte code engine and a compiler to translate SQL into that byte code to eliminate any reliance on an external server process. The original problem involved an arbitrary directed graph that was NP-complete, meaning naive algorithms could not find the optimal valve combinations for damaged pipes. This project began during a funding hiatus caused by a political fight between Newt Gingrich and Bill Clinton. Hipp, Dan Kennedy, and Joe Mistachkin maintain the software today. Motorola approached Hipp in 2001 or 2002 to include SQLite in a new cell phone operating system. Hipp provided a quote of $80,000 for those enhancements.

Ubiquity and Rigorous Testing

SQLite powers everything from mobile phones to aircraft. The engine exists in every Android and iOS device, and Android maintains a 71.94% global market share in 2026. WhatsApp uses local SQLite databases to store messages for 2.5 billion users. Airbus confirms SQLite runs the flight software for the A350 XWB family, and the team promised to support that software through 2050 to meet DO-178B safety requirements. There are over one trillion SQLite databases in active use. The team maintains the code with 100 million lines of test code for 160,000 lines of production code. They achieve 100% branch coverage at the machine code level. This testing rigor follows a year where Hipp worked 60-hour weeks to resolve bugs found on Android devices. The U.S. Library of Congress also recommends SQLite as an archival format for digital preservation. Major users include Facebook, Google, and Apple. The database also runs on smart TVs, automotive multimedia systems, and routers. SQLite remains the most deployed database engine in the world.

Architecture and Performance

The architecture relies on a single file on disk that contains all tables, indices, triggers, and views. This serverless design allows SQLite to run as an embedded part of an application. It remains 35% faster than the filesystem for reading and writing 10KB BLOBs. Because the engine uses file-based locking, only one writer can access the database at a time. If a second writer attempts to write while the first is active, the system returns an SQLITE_BUSY error. This serialization limits SQLite to low-concurrency workloads. At 5ms per write, SQLite can handle 200 writes per second sequentially. You already know how ubiquitous mobile OSs are, so consider how they all rely on this one library. Will the move to a new storage engine eventually replace the B-tree implementation? SQLite files are bit-for-bit identical on 32-bit, 64-bit, big-endian, and little-endian platforms. The library size stays under 600 KB. It supports Python since version 2.5, PHP since 5.0, and Ruby for Rails.