I really like the idea of storing logs in parquet format in S3. It's something I've wanted to play with for a long time, but never have had a use-case. I'm also a big fan of Rust.

I imagine you've seen apenwarr's blog post on logging: https://apenwarr.ca/log/20190216

With that in mind I'd like to read more about the architecture. Specifically:

1. Do you have separate components for ingest and querying that can be scaled independently?

2. How dumb is the log ingest path?