Corey Lovitt wrote:
Corey Lovitt wrote:
Hi,
I've been working with virtuoso for a couple weeks now
and have been able to load a test RDF data file into the
quad store (subset of the UNIPROT dataset containing about
2.6 million triples) and query it with good
success. I had saved the loading commands as well as
SPARQL queries for the database so they were reusable.
In testing the system, I shutdown the virtuoso-t
server, deleted the db-files as well as the trx and
the pxa files and then brought the server back up. I
then did a full reload of the dataset.
I'm now struggling with query performance on what seem
to be very basic queries and am looking for suggestions to
help resolve them. I've loaded the same dataset and
reused the coded SPARQL queries it. My original
install was I believe into a vanilla quadstore (no special
indexes in place). I believe the query is being done
on exclusively local data - if this assumption is wrong it
may explain the performance issue - really don't understand
how the caching aspect of this system work and may be
tripping on this.
Can folks suggest any way to check this (my packet
sniffer isn't showing anything obvious).
I can provide the explain output for an example query
if this would be useful.
Any suggestions would be _really_ appreciated.
Corey
_______________________________________________
Virtuoso-users mailing list
Virtuoso-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/virtuoso-users
Corey,
What indexes do you have in place ?
What RDF data loading guide did you use and follow?
--
Regards,
Kingsley Idehen
Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO OpenLink Software
Web: http://www.openlinksw.com
I'm using the default indexes: 'RDF_QUAD' and 'RDF_QUAD_OGPS'.
The query I'm running uses a FROM to name the graph so I was assuming that
default indexes were appropriate.
PREFIX uni: <http://purl.uniprot.org/core/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
SELECT ?Protein ?ProteinName ?FunctionDescription
FROM <http://localgraphstore/data/UNIPROT>
WHERE {
?Protein uni:organism <http://purl.uniprot.org/taxonomy/9606>. #get Human records
?Protein uni:recommendedName ?recnameid .
?recnameid uni:fullName ?ProteinName .
?Protein uni:annotation ?annotationid .
?annotationid a uni:Function_Annotation ;
rdfs:comment ?FunctionDescription .
}
In terms of loading guide, I used the information available through the
5.0.{9,10} documentation releases. The data was loaded from the isql prompt
using the commands:
checkpoint;
checkpoint_interval(6000);
DB.DBA.RDF_LOAD_RDFXML_MT
(file_to_string_output('/opt/bigV/data-to-load/uniprot-test-records.rdf'),'',
'http://localgraphstore/data/UNIPROT', 0, 1);
checkpoint;
checkpoint_interval(60);
I've just upgraded to 5.0.11 and it looks like some of the docs have been
updated. Performance/tuning section 15.7 - anything new in it?
Cheers,
Corey
Corey,
Please read: http://docs.openlinksw.com/virtuoso/rdfperformancetuning.html
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO
OpenLink Software Web: http://www.openlinksw.com
