Precisely an implied part of my point was, server-side caching, DB replication, CAP, etc are all relatively well-understood problems.

One can solve those without reinventing the database, as the article denounces.