@samnewman ), although quite interesting (and contradictory, and questionable... but sure interesting), added a bit to that "NoSQL being more productive" silliness:
During the Jurassic period when SQL databases did not fit into Personal Computers I did work a lot with NoSQL databases. I sure remember how painful it was and how I got much more productive with the SQL ones.
If you do not think a DSL is needed to make programming data manipulation tasks more productive, just take a look to what happens with Apache Hadoop and Pig Latin:
(*) It is often possible to just keep all the data in memory and serialize it all as one big doc.
And even after this NoSQL-later-years trend started, this discussion, about history repeating and lessons from the past not being learned, is already ongoing at least since 2009 (check referenced articles too):
What is new about the new vague of NoSQL databases, is the way some of them are targeting Big Data and the new strategies they use to deal with redundancy and consistency, especially for high volume + high concurrency scenarios.
Having a query language or not is mostly orthogonal to consistency models - although it might somewhat affect the query language just like it affects the whole database use model.
What I am questioning here is the claim about higher programming productivity of NoSQL databases for those most common cases where the SQL databases or no database alternatives would do.
Remember: most people are neither building another Twitter nor another Google.
So, again, what I am having trouble to understand is:
- How is NoSQL so much more productive for others than the alternatives mentioned above? That completely contradicts my experience.
- Why are query languages (with some SQL likeness) popping up if NoSQL database APIs are the real deal?
- Or how is this silliness still going on since 2009?