A pipelined strategy for processing recursive queries in parallel

摘要

This paper introduces a pipelined strategy for handling recursive queries in parallel. The proposed strategy consists of two steps in processing a recursive query. First, a query is compiled into a canonical form. Then the derived form is evaluated on a generalised pipe. The strategy is potentially applicable to any kind of recursive queries defined over the domain of function free Horn clauses. By pipelining the evaluation process on a parallel architecture, and restricting access to only the data which directly contribute to answering the query, the new strategy attempts to provide high efficiency for processing recursive queries in deductive database systems. A short qualitative comparison with some other approaches is also presented in this paper to show the usefulness of the new strategy.