Data Management

Robust and Efficient Algorithms for Rank Join Evaluation

Date Added: Jul 2009
Format: PDF

In this paper, the authors examines rank join operators by considering both I/O cost and total execution time. Their empirical study of existing state-of-the-art operators showed that they are not effective when evaluated under both metrics. This led to the creation of the FRPA operator which is instance-optimal and far more efficient than the currently known instance-optimal operator. The authors also introduced the a-FRPA operator that can adapt its behavior automatically between the instance-optimal FRPA operator and the empirically efficient HRJN operator depending on the input. Their experimental results validated the effectiveness of the new operators and demonstrated that they offer significant performance improvements compared to existing operators.