You can already use Spark SQL to query files (e.g. CSV, JSON, Parquet) directly on various storage systems without ETL. It doesn't have as much research novelty

You can already use Spark SQL to query files (e.g. CSV, JSON, Parquet) directly on various storage systems without ETL. It doesn't have as much research novelty as NoDB, but it is used in a lot of production data pipelines and ad-hoc analysis on "raw data files".