Show HN: OasysDB, Storing vectors for RAG in Rust simplified

3 points

2 years ago

Hi hackers, this is Edwin and I'm building OasysDB, a SQLite-inspired embedded vector database in Rust. My goal is to simplify the process of storing vector embeddings while still having high performance with simple APIs.

Currently, it's only available in Rust via "cargo add oasysdb" command. But, I'm working to make it interoperable with Python ASAP. If you want to test it out, simply head to the GitHub repository (https://github.com/oasysai/oasysdb) and refer to the Quickstart section.

Performance-wise, I have some quick results that I can show you. I got this result locally using my machine of Apple M2 chip and 16GB of memory.

- For 10,000 vectors with 128 dimensions + usize metadata > Search time: 0.15 ms > Memory usage: 6 MB

- For 1,000,000 vectors with 128 dimensions + usize metadata > Search time: 1.5 ms > Memory usage: 600 MB

I'd love to hear some feedback from the community to help make OasysDB better.

4 comments