Hi, I'm running the SP2 benchmark [http://dbis.informatik.uni-freiburg.de/index.php?project=SP2B] which consists of timing the performance of a number of queries against differing sized sets of triples.
The benchmark comes with a dataset generating tool, which models scientific publications.
I've run into some odd behavior regarding one of the queries:
This is supposed to select all distinct pairs of authors that have published in the same journal, and returns very, very large datasets far exceeding the existing number of triples it is selecting from.
I'm running this query 25 times and the behavior I am seeing is that after a few successful runs, the rest will fail with a network connection error, leading me to believe that Oracle is refusing further connections.