I'm planning to implement a CouchDb database that is likely to have thousands of very small databases. I'm planning on using the standard railway install with volume attachment, but I'm wondering if I need to engineer a way to cluster them out or not (yet) based on the limitations of this structure. Since each database has a separate file (at least that's my understanding), this can cause issues in unix systems with max files in a directory/max files accessed at once. I'm not expecting a ton of databases to all be open at one time, but it might be close to 500ish.

I don't think it should cause issues, but I wanted to ask from an infrastructure standpoint if it's likely to hit resource limits.
This is the documentation page on resource limits from CouchDb: https://docs.couchdb.org/en/stable/maintenance/performance.html

Not sure how this works internally so if someone who's more familiar with this server setup can help me out, that'd be awesome!