- From: Orri Erling <erling@xs4all.nl>
- Date: Thu, 26 Feb 2009 13:15:55 +0100
- To: "'Seaborne, Andy'" <andy.seaborne@hp.com>, "'SPARQL Working Group'" <public-rdf-dawg@w3.org>
Introduction: Orri Erling Role: Secondary contact for OpenLink Software I am the program manager for OpenLink Virtuoso. I participated in the previous DAWG, having joined it in the middle of its term. My background is in databases, from storage engines to SQL optimization. I have lastly written cluster support for version 6 of Virtuoso, to be commercially available during the first quarter of 09. The value proposition is to offer one two two orders of magnitude more scalability for RDF storage plus a SPARQL with sufficient extensions for any typical database workload and more. Many details of this work are discussed at the Virtuoso blog, http://virtuoso.openlinksw.com/blog . Presently, this is being used for hosting a large part of the Linked Open Data cloud, offering full text and SPARQL query and faceted browsing on well over a billion triples. Besides supporting the SPARQL protocol and diverse SQL clients, we are also compatible with the Jena and Sesame frameworks. Work to be undertaken after productization of the present Virtuoso Cluster platform involves dynamic scaling and partitioning in a database cloud environment, federated querying and integrating more inferencing into the database. For the present working group, we find it important to bring SPARQL to parity with SQL in SQL's areas of strength while preserving the advantages of the RDF data model. Besides this, we find that the open web character of RDF warrants specific support for federated queries and discovery. We will shortly publish a wiki page outlining how we propose to accomplish these goals. Most of the needed features are already implemented by ourselves and others. For the working group, we would like to see the maximum ground covered in standardizing these. The creation of a SPARQL recommendation featuring the required expressivity is the aim, not per se the adoption of our solution to any particular issue. I join Andy in wishing to concentrate on the most practical aspects first. This, as well as the community involvement Andy called for, should be facilitated by the existence of live implementations of many of the features we propose to agree upon. Ivan Mikhailov is our primary contact. Matters requiring rapid response should be addressed to him, with cc to myself. Regards Orri Erling
Received on Thursday, 26 February 2009 12:17:02 UTC