Also, the RDBMS world is full of some of the oldest scaling experts in the yellow pages. It's interesting how surprised people were that many of the "traditional" RDBMS were able to catch up on some of the scaling support that were the biggest advertised advantages of NoSQL.

It's interesting because I feel like the NoSQL world spent a lot of time reinventing the RDBMS from the "opposite direction". For all its faults, SQL is a fascinating language because it mostly ignores low level details of how the database scales, how it operates under the hood. It's not that SQL is intrinsically hard to scale (certainly at the relational algebra roots it shouldn't be hard, in theory), but it certainly leaves a lot of work to anyone building a database engine to figure out what/when/why/how to scale. I feel like a lot of RDBMS' query analyzers/planners resemble things like HBase a lot more than folks realize.

It's great that NoSQL realized that sometimes those "low level details" in an SQL engine are useful in their own ways, and have increased the spectrum of performance versus power/flexibility trade-off options. But it shouldn't be that big of a surprise that SQL databases remain competitive in that trade-off space, given under the hood they've had to think about a lot of that stuff over many decades.