Performance recommendations for read heavy app

[mahesh-hegde]

I have a read-heavy application, bottlenecks appear mostly in fuzzy search and DocIDs query (the latter is usually around 50-100 Doc IDs in a bulk retrieval which takes around 10ms each request. I have ~200k entries if that matters).

I can cache them per ID as key, but that won't solve the root cause.

It appears certain index types can be chosen using a config map in OpenUsing, but it doesn't seem to be documented.

Any config recommendations to speed up this use case?