Inclusion of databases?

Howdy Folks,

I see that parallel processing stuff and linear regression is supported to work distributedly
from an input which is a csv file.
I want to execute a clustering algorithm with data which is stored in database. Yeah I know there is a huge downside to actually
just downloading the data of a file, since the query operations might take time (they are simple) and the transmission from server to client even more.
But is there a possibility to actually use as input from rows the databases and whilst caching data (Vectors for examples) in the working memory or/and store the important metrics of the algorithm in database table?

The report names no file, test, or entry point. Start by reviewing existing Dask-ML data-loading and clustering integrations, then clarify whether database-backed input, caching, and metric persistence are all in scope. Completion criteria are not defined in the issue.

Written by the indexing model from the issue text.