Gives you unified access to all your data spread across all your relational databases.
Supports: PostgreSQL, MySQL, SQL server, Oracle and DB2
Allows you to query across all these databases through a unified interface. Supports reasoning to allow for concepts to be modelled differently in each relational database, but be queried through a single model. This means that developers higher up the food chain don't need to know that the old DB2 instance from 2005 that stores experimental turbine data uses a new table per turbine while the new Oracle db uses a single table with column to specify the turbine instead.
Looks interesting. Years ago, I experimented with the D2RQ project that does something similar.
It would be interesting to see a benchmark with something like WikiData or DBPedia converted to a relational database and compared with a RDF store like Virtuoso. Compare query speed, memory use, and index size on disk. Also seeing a comparison of bulk data load times.
The basic idea was done in a nice study by Orri Erling at the time working on virtuoso (now Facebook presto). Which indeed shows the major benefits of recovering a physical schema.
I've always wondered if the Stardog implementation is based on the work done by Ontop? I can see from the history of Ontop that they also have releases dating back to 2014.
Indeed, the first implementation of VKG in Stardog v4 in 2014 shipped with Ontop v1.16. Since then, Ontop has evolved and it improved a lot in terms of features and performance.
7 comments
[ 3.6 ms ] story [ 27.1 ms ] threadSupports: PostgreSQL, MySQL, SQL server, Oracle and DB2
Allows you to query across all these databases through a unified interface. Supports reasoning to allow for concepts to be modelled differently in each relational database, but be queried through a single model. This means that developers higher up the food chain don't need to know that the old DB2 instance from 2005 that stores experimental turbine data uses a new table per turbine while the new Oracle db uses a single table with column to specify the turbine instead.
It would be interesting to see a benchmark with something like WikiData or DBPedia converted to a relational database and compared with a RDF store like Virtuoso. Compare query speed, memory use, and index size on disk. Also seeing a comparison of bulk data load times.
Anyway, looks like a cool project!
http://www.www2015.it/documents/proceedings/proceedings/p864...