Re: Uniprot RDF in RDF Gateway

Geoff Chappell wrote:
> Well, there's no doubt that we're benefiting from caches for the example
> queries (both the Gateway page cache, and the OS disk cache). If I restart
> the Gateway application/server, the queries run as quickly so I suspect the
> disk cache is providing the greatest benefit). That said, no cache would
> make up for a lack of efficient indices on such a large database. The box
> the queries are running in has 4GB of ram and the database is 27GB (of used
> space), so random queries are bound to go outside of any cache - sooner or
> later some query response times will be somewhat limited by the speed of the
> disk. 

That's quite impressive, then. I assume clearing the disk cache would not 
cause queries that executed within a second to suddenly run for minutes... 
Still, I'll have to see for myself before I believe it :-)


> Not yet. I'm planning to add a sparql interface and will probably throw up
> something that lets you query with our native query language (since it's
> more expressive than sparql). I've also done a bit of work on an
> ontology-driven search/browse interface for it which I'll likely complete at
> some point. I'll keep you posted.

That would be great. I have some interesting queries I'd like to try out, 
if possible.

Received on Thursday, 12 May 2005 12:58:52 UTC