Tradeoffs in Processing Complex Join Queries via Hashing in Multiprocessor Database Machines
Donovan A. Schneider, David J. DeWitt · Minds at UW (University of Wisconsin) · 1990
In this paper we examine the problem of processing multi-way join queries (on the order of 10 joins) through hash-based join methods in a shared-nothing database machine. We first discuss how the choice of a format for a complex query can significantly affect performance in a multiprocessor database machine. Several query processing algorithms are then proposed and experimental results obtained from a simulation study are presented to demonstrate the tradeoffs of left-deep and right-deep scheduling strategies for complex join query evaluation. These results demonstrate that right-deep scheduling strategies can provi& significant performance advantages in large multiprocessor database machines under many circumstances, even when memory is limited.