DWAHP: Workload Aware Hybrid Partitioning and Distribution of RDF Data

Abstract

Proliferation of RDF data has reached to a peak where data is partitioned across multiple nodes. Significant contribution for developing solutions to manage RDF data in distributed environment is witnessed in recent years. We propose a workload aware hybrid partitioning approach for a distributed environment. The objective of our approach is reducing query joins and inter-node communication leading it to faster query execution for frequent queries. Our approach considers a query workload and partitions data based on workload information. It distributes data by exploiting underlying structural relationship between properties using a property reachability matrix to optimize query performance. DWAHP gets rid of inter-node communication cost for frequent queries like linear and star queries and answers 83% of frequent query workload without inter-node communication. DWAHP is compared with state-of-the-art solutions in terms of query execution time, query cost, storage space, and inter-node communication. It has demonstrated significant improvement over state-of-the-art solution.