OK, if we’re going to have a significant increase in noSQL approaches to big data management, what will take the place of SQL?
This isn’t just a Zen koan. SQL, a language I never cared for, nevertheless has two signal virtues:
- By being essentially universal (I know, I know, more honored in the breach than in the observance), SQL provided for separation of concerns between the data layer and applications. It at least defined in principle some kind of border even if, like the border between Kenya and Somalia, it’s something of a literary device.
- By separating the results of a query from the procedure for the query, SQL allowed both to be honed separately. We have good storage engines today and good data science because the two are separated.
What will take the place of this border in the noSQL world? Today it’s anarchic: the query is a method in some languages, a specification in others, and (thanks to some bridging technologies), SQL itself.
Just as web applications took UI/UX back a decade, noSQL risks taking the data layer abstraction back a decade or two. Needs some work.
Anybody know good companies or approaches to this problem?