Matthias Endler

We tried to migrate that to containers and it wasn't really easy because...

operationally java had some unfortunate default decisions where it would try to allocate as much memory as it could possibly get on the machine and if it's co-hosted with a lot of other services on the same machine that sometimes caused some problems so did you also run into or did you also consider these kind of concerns the operational part of it the deployment process

I see.

That's an aspect that I never heard before.

Because Rust is so fast, you can afford to simplify operations by doing more at startup, but it's still performing.

So you can cut some corners thanks to Rust.

What do you use for the storage layer?

And we link to it in the show notes.

It's a really, really nice write-up.

RocksDB is very fast and very write-focused, but also it's effective in terms of storage.

And I guess that plays in your favor because if you try to geocode the world, you need a lot of storage.

And on top of it, once the storage is quite optimized, you get really decent cache locality on top of it for free.

So it might have been a really great choice.

And all of that combines into a very efficient single binary lookup service.

What do you use for the fuzzy searching part that you mentioned?

Yeah, exactly.

Jonathan Gallant, yeah.

Yeah.

The way I always think of an FST, and I might be wrong here, is it's a very fast state machine for looking up the existence of words very efficiently and very quickly.

So you mentioned the tri-data structure that is similar where