Touil Sami wrote:
> Hi,
>
> I just discovered Simile project and i am really interrested in
> Longwell in order to browse data from an application i'm working on.
> This application is really heavy, but the datas are in RDF format ,
> and Longwell could be a perfect system to browse it on remote
> platforms.
>
> But the datasets i'll be using are really heavy too ( around 15 000
> 000 records ) so i was wondering if Longwell was able to manage such
> datas, and which repository ( Jena, Joseki, Kowari, 3store or Sesame )
> i should use.
We have been testing longwell1 with several million statements and an
in-memory jena store and the performance on a normal machine is
outstanding, a 32 bit machine will hit a 4Gb memory limit, which is very
likely going to stop you if you have 15M records and each of them has at
least a few statements.
Joseki is not a repository, but a web interface for a jena store.
Kowari, 3store and Sesame all work reasonably well but we have not (yet)
did extensive profiling on them with a lot of data. It's planned for the
next phase of our project. For the queries that we tried on 300k
triples, we did not observe substantial differences between the various
ones. See Ryan's report for more information.
http://simile.mit.edu/reports/stores/
DISCLAIMER: Ryan's tests were done on a laptop and were just enough for
us to understand if some triple store was dramatically better than
another. We don't claim that report to be extensive or even up-to-date
with the latest releases of the various stores.
We are definately not interested in a 'my tstore is better than yours'
kind of arguments, but we also understand the need to know which one is
going to stand the pressure that you apply.
At this point, no serious and in-depth tstore report exists. One is much
needed, IMO, and SIMILE is going to try to come up with one in the next
6 months, hoping to be as exhaustive as possible (and maybe also help
the tstore implementors to fix problems they might not know they have).
For now, pick one, load things and see it if works. If not, complain
here first.
--
Stefano Mazzocchi
Research Scientist Digital Libraries Research Group
Massachusetts Institute of Technology location: E25-131C
77 Massachusetts Ave telephone: +1 (617) 253-1096
Cambridge, MA 02139-4307 email: stefanom at mit . edu
-------------------------------------------------------------------
Received on Tue Sep 06 2005 - 21:14:09 EDT