Scalable SAPRQL Querying Processing on Large RDF Data in Cloud Computing Environment

Wu, Buwen; Jin, Hai; Yuan, Pingpeng

doi:10.1007/978-3-642-37015-1_55

Cited by 4 publications

(5 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results from the two subqueries are joined in a MapReduce job. A similar approach is followed in [90], where subqueries are processed sequentially (one job per subquery) and then are joined in a left-deep tree manner (one job per join).…”

Section: Join Evaluationmentioning

confidence: 98%

“…In [93], either the repartition or the broadcast join is used, depending on whether the property file is small enough to fit into memory. Finally, in [93] and [90], some pruning techniques are used to reduce the number of intermediary results that are shuffled in the network. In the former, the authors use bloom filters on the subjects or objects of small property files and in the latter the minimum and maximum values of each variable in a job are stored to be used in the subsequent jobs as a filter condition in the SQL query.…”

Section: Join Evaluationmentioning

confidence: 99%

“…Along the data storage dimension, we identify: (i) centralized RDF stores, (ii) key-value stores, and (iii) distributed file systems. With respect to query processing, we distinguish: (i) processing queries locally at one or multiple nodes (without inter-node communications), which supposes that each node has the corresponding RDF query evaluation capabilities; (ii) implementing a fully distributed evaluation engine on top of MapReduce, (iii) implementing a fully distributed evaluation engine on top of a parallel processing Trinity.RDF [91] WARP [42] Partout [37] Huang et al [43] Wu et al [90] MAPSIN [76] H 2 RDF [65] SHARD [72] HadoopRDF [45,44] RAPID+ [70,50,51] PigSPARQL [75] EAGRE [92] Zhang et al [93] Huang et al [43] WARP [42] AMADA [21,11,22] H 2 RDF [65] Rya [68] Statustore [78] CumulusRDF [54] AMADA [21,11,22] AMADA [21,11,22] Key-value stores DFS Centralized RDFstore/RDBMS Locally MapReduce…”

Section: Qualitative Comparisonmentioning

confidence: 99%

“…Finally, in [90], SPARQL queries are also decomposed into subqueries, based on the data partitioning. The subqueries are translated into SQL queries and sent for evaluation to the underlying RDBMS of HadoopDB [4].…”

Section: S P O) Dl(*) Dl(*)+csel Dl(*)+csel Dl(*)+ssel Dl(*)+csel Dmentioning

confidence: 99%

“…Finally, in [90], HadoopDB [4] is used, where RDF triples are stored in a DBMS at each node. Triples are stored using the vertical partitioning proposed in [1], i.e., one table with two columns per property.…”

Section: Federating Multiple Centralized Rdf Storesmentioning

confidence: 99%

See 4 more Smart Citations

RDF in the clouds: a survey

2014

View full text Add to dashboard Cite

The Resource Description Framework (RDF) pioneered by the W3C is increasingly being adopted to model data in a variety of scenarios, in particular data to be published or exchanged on the Web. Managing large volumes of RDF data is challenging, due to the sheer size, the heterogeneity, and the further complexity brought by RDF reasoning. To tackle the size challenge, distributed storage architectures are required. Cloud computing is an emerging paradigm massively adopted in many applications for the scalability, faulttolerance, and elasticity feature it provides, enabling the easy deployment of distributed and parallel architectures. In this article, we survey RDF data management architectures and systems designed for a cloud environment, and more generally, those large-scale RDF data management systems that can be easily deployed therein. We first give the necessary background, then describe the existing systems and proposals in this area, and classify them according to dimensions related to their capabilities and implementation techniques. The survey ends with a discussion of open problems and perspectives.

show abstract

Section: Join Evaluationmentioning

confidence: 98%