The Research Of Distributed RDF Data Processing Architecture

RDF(resource description framework)describes resources and the relationships of them on the semantic Web,which has been widely used in life.With the volume of RDF data rapidly increasing,a high performance method is necessary to efficiently processing SPAQRL(simple protocol and RDF query language)query over RDF data.This paper analyzes a large number of distributed RDF data processing engine,and based on the analysis results proposed query performance optimization strategy.We survey current parallel processing approaches for SPARQL queries in RDF data.This paper research the engines from three aspects: system architectures,RDF data storage management,and SPARQL query executive strategies.We classify those architectures,respectively.Combining the above analyze,most existing approaches have been proposed to processing RDF data,which mostly depend on calculative capabilities of CPU.In recent years,GPU(graphic processing units)has been adopted to accelerate graph data processing widely in several works,which have better computational performance,superior scalability,and more reasonable prices.Due to the limited computational power of CPUs for handling large-scale RDF data,this article considers the use of GPUs as a complement to system computing power.The system includes query splitter,GPU-based MapReduce parallel computing templates and computing CPU and GPU resource management to reduce the response time of SPAQRL queries.And the performance of the algorithm testes on the benchmark data set LUBM.The experimental results show that the algorithm is superior to the algorithm based on CPU architecture.To sum up,this paper designs a distributed RDF data query system based on MapReduce GPU for the current situation of low query efficiency.The system through new hardware acceleration,greatly improved system performance.A new attempt was made to use hardware acceleration in the RDF data processing system.