What's new is old again. Michael Stonebraker had a paper about this in 2007 [0]. Over that time a slew of products came into the market that claimed to do it all. JSON, SQL, ACID transactions, Streaming, time-series, Full-text search, batch processing, ETL etc. Things that don't even fit in the same bucket. For awhile companies were positioning Hadoop to do all of it
Seems like through every technology hype cycle, the same reminders need to be written and re-circulated again.