LAM

Location aware SPARQL query implementation over GraphX

Arguments to supply (Local mode):

-g “/path/to/N3File.nt” -q /path/to/queryFile -o /path/to/outputFile -p 4 -m local

where

• g is the path to the RDF file in N3 format stored in HDFS
• q is the local path to a query file
• o is the hdfs path to output
• p is the level of parallelism desired/ number of spark cores (here set to 4)
• m is optional and needs to be set as “local” when only run locally

Arguments to supply (Cluster YARN mode):

spark-submit –class main –master yarn –deploy-mode cluster –num-executors 6 –executor-cores 4 –files qyeryFile LAM.jar -g “/path/to/N3File.nt” -q /path/to/queryFile -o /path/to/outputFile -p 4

• num-executors is the number of machines to be used in the cluster (here 6)
• executor-cores is the number of cores per machine, set it equal to the level of parallelism desired (here 4)