Right.. Databricks is a platform to perform transformations.. Ideally your should either mount s3 bucket or ADLS gen 2 location in DBFS..

Read/Write/Update/Delete your data and to run SQL analytics from SQL tab, you'll have to register a table and start an endpoint..

You can also query the data via notebooks by using SQL in a cell. The only difference is, you'll have to spin up a cluster instead of an endpoint.