Set-Oriented Memory Management In A Multiprocessor Database Machine

In the past few years research in the area of database machines has been centered mainly around efficient query processing and increased I/O-bandwidth. A large variety of software and hardware mechanisms has been proposed, including various forms of filtering techniques and parallel query evaluation in multiprocessor systems. However, the issue of memory management under these new techniques has received only little attention. In this paper we analyze the effects of employing filtering techniques and set-oriented query evaluation in multiprocessor systems on memory management. We propose a set-oriented memory management method which has been designed for a multiprocessor database machine utilizing software filtering and parallel query processing under dataflow control. We also address the problem of performing updates in a set oriented environment.