David Webb-CDS (8/6/2014)
The other thing you are leaving out is the vendor ecology. RDBMS technology and SQL had great success partly because it provided a consistent data manipulation method across multiple products. All the report writer vendors and analytics vendors didn't have to tailor their products for 15 different languages and data storage mechanisms. If you have to write a generic tool for 15 different No-SQL languages and 15 different data storage paradigms, it costs more and your tool will cost more.
I'm pretty sure you are right there.
But we do need a replacement for SQL; IBM screwed up badly when they pushed Ted Codd out of the way and redesigned his relational calculus wheel to have several very sharp corners and the odd completely flat or even concave section of outer surface here and there. The trouble is that now, about 4 decades on, everone has been living with the faults of SQL and has got used to them, and fixing SQL would be a nightmare - it was already just about impossible to get anything non-trivial changed in the SQL standard 25 years ago .- so we need a new language and we need all the RDBMS suppliers to commit to offering it if it is not to be a waste of time. So we may have to do without.
Tom