Added query optimization to GraphSailConnection

[niclashoyer]

I ran into some performance problems when using the Sesame ouplementation and figured out that there was no query optimization.
I added basic query optimization that should speed up query execution in most cases.

[spmallette]

@joshsh can you please take a look at this one when you get a chance?

[joshsh]

@niclashoyer I'm happy to merge this if you can demonstrate a performance boost for your queries. I gave some similar optimization a try, a while back, but removed it, as it seemed to have little impact: 3f2b225

[niclashoyer]

I'm developing a SPARQL-plugin for Neo4j that'll be released in the next months. At the moment I'm benchmarking performance of the implementation against other graph stores (such as Fuseki).

I'm using the Berlin SPARQL Benchmark (BSBM) to perform these tests. Basically it generates a dataset of a desired size and then it runs different queries against the graph store and records the performance.

I already had a discussion on the Sesame mailing list about some queries that perform bad, because reordering them manually improved the performance significantly.

To verify that the query optimization improves performance I set up a small dataset (8498 triples) and ran the tests with and without query optimization. Without optimization the implementation runs 2365.05 query mixes per hour (each query mix consisting of several queries). With the optimization it runs 13068.70 query mixes per hour. For detailed logs see this gist.

The difference gets worse if the dataset grows.

[niclashoyer]

@joshsh is that enough information for the query optimization? I could probably do some more tests, but that would take some time.