Hardware-Conscious DBMS Architecture for Data-Intensive Applications

Recent research shows that traditional database technology does not efficiently exploit the features of modern computer hardware. This problem is especially visible in the area of computationally-intensive applications, where specialized programs achieve performance orders of magnitude higher than DBMSs. In this article we present techniques that aim at bridging this gap. Three issues are discussed: efficient query execution on superscalar CPUs, optimizing disk bandwidth on commodity hardware, and exploiting the features of emerging multi-threaded and multi-core CPUs.