Hi!
Ian Wilson wrote:
> Great. I'm looking at large graph in memory storage and query
> performance (SPARQL when available). I looked at the existing Jena and
> Sesame 1.2.2 work in CVS - thanks for the reference. Any thoughts to
> test Sesame 2 and the new SPARQL interface? Or, Kowari? I am not
> familiar with 'cholesterol', but am open to other suggestions. It's been
> about a year since I've looked at Jena too.
With regard to performance, we had a design meeting today for our
nextgen product stuff and I suggested that we use RDF as the internal
engine for pretty much *all* our upcoming projects. Because it made
enormous amounts of sense, and other people on the team had come to the
same conclusion independently of the research I had done lately.
Basically we want to put document (OpenOffice), calendar
(GroupWise+iCal+Exchange to begin with), forum, news (RSSx.x+Atom),
workflow (etc.) data into it and then create massive numbers of views to
be shown in a portal setting, along with replication using P2P via JXTA
(or somesuch) to allow for clustering and replication of data. From what
I can see of this is pretty trivial except for the performance and
scalability part.
What RDF backends support diskbased storage, and are any of the ones
backed by RDBMS's any good? Specifically, if an RDBMS is used we'd like
to use Apache Derby if possible, because of cross-platform concerns.
What are the largest RDF-systems today, and are there any performance
studies done? What is the general "feeling" among RDF-people with regard
to what kind of performance is possible? (hunches is fine, we're not
talking exact sciences here :-)
Any experience or suggestions around these areas is welcome.
/Rickard
--
Rickard Öberg
rickard.oberg_at_senselogic.se
_at_work +46-(0)19-173036
Received on Mon Nov 14 2005 - 14:37:56 EST