A major point made in the article is that the standard big data stack does not fit all big data needs.

There's a growing assumption that this stack is sort of all that's needed, and that's just not the case.