With the move to the "edge" for compute, having your database at the edge reduces latency and basically makes it viable. So the theory is that SQLite read replicas will be the answer to a lot of these plans. See what Fly and Cloudflare are investing in with the edge network and SQLite plans.

On top of that SQLite is now viable in the browser with WASM and the "origin privet file system" api. And so people are increasing looking at moving more state and persistence to the browser with an eventually consistent sync. This is what the "Local First" movement, myself included, are excited about.

Sqlite is extremely powerful for any locally-embedded databases, such as apps with an offline component. For example, searching text in notes, books, cached news, etc. It's very simple to implement inside Sqlite with much less overhead than including a Postgres/Mysql instance.

Lol, you’re using it “in production” right now! SQLite is everywhere, almost definitely in the browser you’re using to view this post.

It is used in a huge number of production environments, including within your browser (most probably).

Probably not something you want to run a multi tenant SaaS database with, but it is useful if you are going the one tenant per database route.

HN has fads like everything else. If this were 10 years ago they'd be asking about MongoDB. 5 years ago they'd be asking if the Blockchain could power their search engine. Now they want an AI search engine.

It’s used in production quite a bit, probably just not in your field as much. Anytime I’m thinking about storing something in a file I use SQLite instead

SQLite has been the default store for CoreData in iPhone apps for years afaik. I imagine that's a lot of apps.

Of course, CoreData sucks, but that's for unrelated reasons.

We do natural language processing with some custom tokenisation, the data it uses at runtime is put into an sqlite db at build-time. Works fine for our usecase.

Put simply - SQLite is incredibly powerful and simple - it can handle a lot more then people might think - and AFAIK it is the most "used in production" database whether you're aware of it or not.

Depending on the use case it can scale incredibly well, and is tiny and battle hardened.

It's the most deployed database, it's on every smartphone, "smart device", computer, it's inside of many of your apps, powers many of the websites you use, etc.

---

tl;dr - If you're following "use the simplest, but reliable tool for the job" then sqlite is a valid option for a crapload of use cases, most people aren't aware of what it can actually do because it doesn't "web scale".

Sqlite isn't new, it's old, and it's "used in production" count is in the literal billions.

I put together a demo of this for work where I pulled all our private repos and built an index, then used datasette as the front end.

It was incredibly fast, so much so that I found it more useful than github search. The index is pretty out of date now but I still use it purely based on convenience and speed