Database slicing on relational databases

Abstract

Many software systems today use databases to permanently store their data. Testing, bug finding and migration are complex problems in the case of databases that contain many records. Here, our method can speed up these processes if we can select a smaller piece of the database (called a slice) that contains all of the records belonging to the slicing criterion. The slicing criterion might be, for example, a record which gives rise to a bug in the program. Database slicing seeks to select all the records belonging to a specific slicing criterion. Here, we introduce the concept of database slicing and describe the algorithms and data structures necessary for slicing a given database. We define the Table-based and the Record-based slicing algorithms and we empirically evaluate these methods in two scenarios by applying the slicing to the database of a real-life application and to random generated database content.