Replacing a 3 GB SQLite DB with a 10 MB FST Binary: Slim Down or Pipe Dream?
Someone shrunk a 3 GB SQLite database to 10 MB using FST — no magic, just finite state transducers.

Imagine walking into work and finding your 3 GB SQLite file has suddenly slimmed down to 10 MB. No, it's not a new database detox trend — it's an experiment with FST (finite state transducer).
How It Works
FST is like a smart combination lock: it stores data as states and transitions, ditching everything unnecessary. Unlike SQLite, which drags along indexes, logs, and other bloat, FST compresses info down to the bare minimum while keeping search speed.
Who Needs This?
- If you have read-only data (dictionaries, geodata, routes).
- If every megabyte counts (mobile apps, IoT).
- If you're a minimalist who hates bloated databases.
Caveats
Of course, FST doesn't do JOINs or transactions — it's not a SQLite replacement, more like a Swiss Army knife for specific tasks. But when you need to store millions of records and search them fast, FST looks like a hero from Inception: compact and effective.
METABYTE studio's take: We love optimization too, but we're not ready to replace all databases with FST — our CI/CD wouldn't survive it. Still, for static data, it's a solid choice!
NEXT STEP
Liked the approach?
We apply the same principles to client projects: AI, automation, products that don't die after launch.