Re: RDF DB Shootout: Preliminary Results

From: Dean Allemang <dino_at_happydino.com>
Date: Mon, 09 Jan 2006 13:29:47 -0500

Vineet Sinha wrote:

> Dean Allemang wrote:
>
>>
>> It is too bad you weren't able to get Redland running in your test
>> harness; completely unscientific and anecdotal evidence we have from
>> a very large data import task indicates that it runs roughly 10x
>> faster than Sesame (in the particular anecdotal situation, the
>> dataset size was so large that this 10x meant the difference between
>> an overnight load, and one that would take a processer running full
>> tilt for a week; a rather daunting proposition).
>> I would love to see the results of a controlled experiment on this.
>
>
> Dean,
> If you have experience with Redland's Java API's then it should not be
> hard to write a driver for it. Let me know if you want to work on this.
>
> Regards,
> Vineet
>
>
>
>


This discussion is getting beyond my competence; I don't have experience
with the API at all. We used Redland as wrapped in Intellidimension's
RDF Gateway. A client was trying to read in a very large data set,
first into Oracle 10g, then into Sesame (Sesame beat 10g by about 10 to
1 as well - or was it 100 to 1? I'll have to look that up), but even
wiht Sesame, his Java programmers were having very little joy.

On a lark, I said, "Try importing it into RDF Gateway [which is built on
Redland]. Yes, you are using a (proprietary) scripting language to do
it, but it won't take you more than an hour to learn how to do it and
try it".

They immediately experienced considerable speed-up over their best
so-far Sesame try. After an afternoon in discussion with Geoff at
Intellidimension, they got the 10x result.

So, to be a tiny bit more scientific (but no less anecdotal), it was
actually Redland + RDF Gateway scripting interface to their API v.
Sesame being used by Java programmers in Java (who know the details).

Dean
Received on Mon Jan 09 2006 - 18:29:44 EST