Generally speaking if you re looking at
Generally speaking if you're looking at firebase and all that I don't think it matters much if you just use postgres like a nosql database aha.
I doubt the scale you'd be looking at would be enough to suitable strain postgres before something like firebase would be unrealistically expensive aha.
However, you can self host supabase with something like yugabyte db which can support fat rows.
4 Replies
I'm not using Firebase for its scalability reasons.
I'm more comfortable using NoSQL than SQL.
And the current project don't have that much relational data
My recommendation for NoSQL databases tends to be ScyllaDB for everything other than analytical type stuff in which case ClickHouse is king
Thanks for your suggestions 🙂
For now, I'm gonna stick with my choices.
I really don't want to learn a new thing rn 🙃
btw, I saw the lust project in your profile.
You got a guide on how to optimize images stored in Cloudflare r2 using this lib?
The system is stateless, just needs a storage backend which can be R2. The easiest way is probably to run it ad a preprocessor with the
aot mode, so that way the images are converted and sized and then stored individually in R2, then you can just serve them directly out of R2/S3/other without needing another server to proxy it like you would the jit/realtime mode
You would likely need to go through all the images you already have stored and run them through lust so it can organize them how it wants.
Worth noting though that lust doesn't currently run in wasm so you can't run it via workers