Processing M-trees with parallel resources

Abstract

The problem of the design and implementation of parallel metric tree indexes, called M-trees, is elaborated. Four different object declustering techniques are proposed and tested in order to get a sufficient evidence needed for specifying the pros and cons of their application. In general, the obtained I/O speedup and scaleup levels are high. A method for dealing with the CPU parallelism is also proposed and its speedup and scaleup experimentally tested.