I feel like this journalist was struggling to understand the nature of different kinds of databases. > document-oriented databases (where different types of da

I feel like this journalist was struggling to understand the nature of different kinds of databases. > document-oriented databases (where different types of data are stored in one unified document instead of a bunch of tabular cells) There isn't one document, and typically documents aren't "unified" in the sense of having an enforced schema. > columnar databases (where data is laid out in key-valued arrays) Er, no. Where data is laid out in columns, to speed up aggregate and analytical operations. > (Confusingly, the NoSQL name corresponds to "not only SQL" rather than "no SQL.") Backronyms ahoy! > But the thing about abstraction in computer science, whether it's a high-level virtual machine-based programming language like Java or a bare-bones Unix shell, is that it always has a cost. So does hard-baking a single storage and access pattern, which is that you get it wrong and have catastrophically slow access and aggregation times. A relational database is a generalist, query optimisers are generally good at guessing how best to traverse a normalised schema and the last time people used "abstractions are too costly!" as a serious argument it was in the days of rewriting desktop apps in assembler.