One thing I find interesting is that there are very few "universal" tools for document databases like there are for RDBMS, even though the APIs for these document databases overlap substantially.<p>For RDBMS, for instance, you have APIs such as ODBC and JDBC that let you connect to different products. There also are tools like Datagrip, as well as tools that can do unison queries against multiple databases (e.g. JBoss Data Virtualization)<p>The basic API of mongodb, elasticsearch, arangodb, couchbase, cassandra, etc. are almost the same when it comes to CRUD operations. Queries are different between those databases but I'm pretty sure you could cross compile AQL and N1QL most of the time, make an abstraction layer that lets you do mongo-style queries against arango, etc.<p>Another factor is that some of the "post-relational" database types are still relational. A columnar database can be perfectly relational, it's just another way to implement it. I've seen some analytics vendors that just don't get document databases, SPARQL, etc. because they can't see past the pure speed of SQL with a columnar internal representation.
tl;dr "The RDBMS is not so much dead as rebranded."<p>This is one of those clickbait-by-arguing-semantics articles. If you follow databases, you won't find anything new or interesting here.