Is your feature request related to a problem? Please describe.
Current semantic caching uses an in-memory struct for storage. More persistent vectorDB should be supported.

Describe the solution you'd like
A pluggable interface so that vectorDB from different providers can be added easily.