At first, it may seem a bit strange to talk about dataset persistence within a data access component library. One might think data is in the database and you simply need to retrieve it, use it, and save it back. But there are a number of situations where these three basic steps get something in between that somehow interrupts this very simple cycle.
Nowadays, when systems are often distributed and your software solution may consist of some client applications (desktop, mobile, or web), one or more application servers (usually REST), and some final storage (still usually an RDBMS), you may find yourself in need of managing a dataset's life in not-so-classical ways.
Let's think about a client mobile application that retrieves data from the server and then has to work out of reach of the server itself. The chance to keep the dataset in memory is gone since mobile applications have their own lifecycle, usually managed by operating systems. Manually...