How to Design an Embedded RDBMS Search

As the cost of micro-disk and NAND Flash continue to drop, devices are storing more and more data. It is common now for a person's MP3 player to have more storage than their laptop. But this increase in storage capacity has not been matched with advances in the user interface. Typically, users still wrestle with a folder-based interface to find the data they want, searching by a few vendor-defined categories such as artist, album, and genre. But a new class of embedded database management systems (DBMS) is emerging to allow end users to search the way people think, rather than in this static manner. With a RAM footprint ranging from a few tens to a few hundred kilobytes, these products enable developers to deliver this sophisticated search on mobile devices. So how do they work? How do you write an embedded application to use a relational DBMS (RDBMS)? While there are a few kinds of DBMS, the relational model has triumphed over all the others, largely because it abstracts the data structures so that applications don't have to know them. A relational database management system offers a standard, high-level query language that allows access to data by content, not by pointer or location and offset.