We have previously discussed RDB2RDF technology and found some information about available implementations. Now we have further explored some of the implementation options. In our survey, we found that there are basically two viable production-level tools for making SPARQL queries against a relational database: the Virtuoso server and the D2R server.
Virtuoso is a very powerful tool for linked data deployment that is available in an open source variety and a proprietary variety. Although the open source version has lots of features, unfortunately the mechanism for connecting to an external database is part of the proprietary version. This means updating our SPARQL endpoint would involve a data dump step an approach that was used in the original zitgist MusicBrainz mappings but something we want to avoid in the LinkedBrainz project.
This leaves us with D2R server which uses the D2RQ mapping language and was the tool used in the DBTune MusicBrainz mappings. To better suit our needs, we've created a fork of D2R server and D2RQ on github. This also includes our mapping file for mapping MusicBrainz NGS database tables to RDF. An experimental endpoint based on this mapping and D2R is available, however this is very much a work in progress. As our mapping becomes more complete, we will shift focus to measuring database loads and optimization. Any feedback is most welcome.