This is really good news IMHO. Having more open, RDF-based triplestore implementations is good for the semantic and linked-data space. Rya is based on RDF4J and runs atop Accumulo which are both pretty well supported projects.