Sliced column-store (SCS): Ontological foundations and practical implications

Abstract

Advances in business intelligence systems based on processing large data volumes are driving efforts toward read-optimized databases. Recently, the use of column-store approaches as a solution for such databases has become quite popular. The main idea behind the column-store approach is reducing I/O requirements through vertical partitioning of data in which only those attributes that are required to answer a query are read. This paper offers two contributions to column-store data models. First, we show that such models can be grounded in ontological foundations that provide a theoretical basis for column-store databases based on representational adequacy. Second, we use these ontological foundations as the basis to propose an extended model of the column-store model called Sliced Column Store (SCS), and show that this model outperforms column-store models for read-oriented queries. © 2012 Springer-Verlag.