Automatic Full Parallel Processing of Hierarchical SQL Queries

ith the processing power of single-core chips running out of steam, hardware manufacturers have turned to multicore processors to make up for the lack of processing power increases in single-core chips. Unfortunately, developing multicore-capable applications is not a transparent or plug-and-play solution. Such applications require developers to use parallel programming techniques—which are not simple and require program design changes. On the plus side, increased funding for parallel processing research may lead to improved parallel processing design, programming, and testing techniques. This article explores an alternative approach that can automatically perform efficient and effective parallel processing of hierarchical data structures.

Before the advent of relational technology, hierarchical structures and their processing were used heavily in applications and databases. However, at that time, hierarchical structures were rigid, so they lost ground to the more flexible relational technology. But because of XML's ubiquity and flexibility, hierarchical structures are making a comeback. Hierarchical structures are both well-behaved and offer advanced processing capabilities. They also offer one unexplored capability examined here: Hierarchical structures and processing are fully and naturally parallelizable.