Hi, I have been doing some preliminary testing of Virtuoso and ran into an very odd behavior. I loaded our test data set of 500K triples using the rdf_sink mechanism into one graph. I also updated the RDF_QUAD table definition to coincide with the performance tuning tips described in http://docs.openlinksw.com/virtuoso/rdfperformancetuning.html . Below is a table with the timing results of different access methods. As you can see, using the SPARQL end point is by far the fasted access method. Does this data match your performance expectations for the different access methods or is there a particular things to watch out for using the JDBC interface?
Thanks, Marco Query Rows ISQL (ms) JDBC (ms) SPARQL HTTP Endpoint hypothesesTriplesWithoutAuthors 2942 1409 6615 93 hypothesesTriplesWithOptionalAuthors 49317 46644 172072 4349 hypothesesTriplesWithUnionAuthors 4144 2267 9463 153 hypothesesTriplesWithTitleFilter 399 175 945 26 hypothesesTriplesWithTitleFiterAndUnionAuthors 11563 5525 23123 428 marchesisHypothesisWithUnionAuthors 52 49 154 12 marchesiHypothesisContainedResearchStatements 787 379 1970 26
