Add a persistent reflection cache option about sqlalchemy-redshift HOT 2 OPEN

This would be really helpful for postgres as well. It's currently taking 4+ seconds to reflect the information for the three tables in my database.

from sqlalchemy-redshift.

I am also really observing performance issues based on reflection time.

It also seems that a near-term optimization would be to limit the information obtained by a query to only what is specified via *args and **kwargs, so that the cache and the queries are kept in better sync in the first place.

Of course, I definitely understand this approach -- it is much more performant for cases where we would otherwise iteratively issue a large number of queries, but that may not be the case; I think having more cache control and/or more control over how much metadata to prefetch would be really useful.

from sqlalchemy-redshift.