Parallel query support for multidimensional data: Intra-object parallelism

Abstract

Intra-query parallelism is a well-established mechanism for achieving high performance in (object-) relational database systems. However, the methods have yet not been applied to the upcoming field of multidimensional array databases. Specific properties of multidimensional array data require new parallel algorithms. This paper focuses on the parallel execution of expensive multidimensional operations on a single very large multidimensional object. Experiences of the ESTEDI project have shown that this scenario is very typical for the database-supported analysis of large multidimensional data. In the paper we concentrate on data parallelism and investigate multidimensional operations concerning their capability to be executed on parts of a multidimensional object. Furthermore, we introduce different possibilities of splitting multidimensional objects for parallel processing and merging the results, which affects the architecture of the parallel query execution. Intra-operator parallelism was implemented in the Array DBMS RasDaMan. © Springer-Verlag Berlin Heidelberg 2003.