While looking for new approaches for quickly affixing a SPARQL endpoint to a relational database, I re-discovered D2R Server and D2RQ API.
The oft-cited DBpedia.org project and Henry Story's Roller Blog Demo both make use of the D2RQ API and the server. I'd recommend that if you want to find out more, start with Henry's SPARQLing Roller document.
To quickly summarize, though: the D2R library enables Jena, Joseki, and Sesame to access MySQL data as if 't were RDF. Much as it is done in ORM, a map file is used to connect the relational schemas and the corresponding ontologies. Appropriately, the map file itself is an RDF document:
D2RQ V0.5 Specification
RDF Schema for D2RQ
N3 Example of a D2RQ Map.
Delving further, I found an offshoot with a project name that was inevitable (in retrospect) - R2D2, a PHP implementation that uses a D2RQ map to access MySQL through RAP.
Are there any posts out there describing the performance of the D2R Server?