Architecture for a massively parallel database machine

Abstract The linear-throughout Semantic Database Machine (LSDM) is an attempt to bring a massively parallel database machine closer to realization. The primary approach undertaken is to alleviate the traditional processor/disk split which leads to inefficiency by building a more homogeneous database system in which processors and disks are combined together. The LSDM is based on a distributed system of many processors, most of them equipped with a local cache memory and a small disk drive. The individual processors are connected by means of a hypercube network topology. The database model employed for the proposed database computer is the Semantic Binary Model.