From: Leo S. <leo...@df...> - 2009-02-03 12:14:31
|
Hi Enrico, Chris, Sesame, Aperture. Today I ran TPTP profiler on a filled NEPOMUK (my nepomuk) that has a lucene index of 800mb. Saving changes to a RDF resource take forever, I had to profile this, my experience is documented [1] The results are written shortly in this ticket: http://dev.nepomuk.semanticdesktop.org/ticket/730 called inside LuceneIndex.addRemoveStatements cummulative time (seconds): * IndexWriter.updateDocument 554s * LuceneIndex.getDocument 264s (calling IndexReader.document(int) with 254s) * IndexWriter.flush 173s Solutions: * to change our rdf->lucene document mapping, perhaps rethinking context? * reduce number of calls to flush? * new lucene? Enrico - does the latest lucenesail maybe fix this? Anyone else working on this also? Anyone also desperate for a fix because your lucenesail-based application is grinding to a halt? best Leo [1] http://dev.nepomuk.semanticdesktop.org/wiki/JavaProfiling -- ____________________________________________________ DI Leo Sauermann http://www.dfki.de/~sauermann Deutsches Forschungszentrum fuer Kuenstliche Intelligenz DFKI GmbH Trippstadter Strasse 122 P.O. Box 2080 Fon: +49 631 20575-116 D-67663 Kaiserslautern Fax: +49 631 20575-102 Germany Mail: leo...@df... Geschaeftsfuehrung: Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender) Dr. Walter Olthoff Vorsitzender des Aufsichtsrats: Prof. Dr. h.c. Hans A. Aukes Amtsgericht Kaiserslautern, HRB 2313 ____________________________________________________ |