You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jackrabbit.apache.org by Alexander Klimetschek <ak...@day.com> on 2009/09/29 14:00:45 UTC
Re: Question: Jackrabbit on Oracle - oracle indexing rather than
lucene's?
On Tue, Sep 29, 2009 at 13:09, David Wagener <Da...@awi.de> wrote:
> But when I compare the performance of both solutions, the search module
> based on “Bundle File-System” is faster than running search module on
> “Bundle Database PM”.
The (lucene) search index implementation is independent from the
persistence manager. Search times should have the same speed if the
content and the queries are the same.
The only difference could be in the time it takes to index data, as
that depends on the latency of the pm.
> My Question: What about storing data in oracle? is there a possibility to
> use oracle indexing rather than lucene's?
Not easily. The search index implementation is made purely for JCR, to
support property and full-text indexing plus all the special features
as described in the spec. And Lucene, especially if the index is kept
locally as now, is faster as an Oracle based index... at least I would
guess so.
Oracle itself offers a JCR binding (they were part of the standards
committee), but only for their XML db AFAIK and I don't know if they
support the full JCR search features.
Regards,
Alex
--
Alexander Klimetschek
alexander.klimetschek@day.com
Re: Question: Jackrabbit on Oracle - oracle indexing rather than
lucene's?
Posted by Marcel Reutegger <ma...@gmx.net>.
On Tue, Sep 29, 2009 at 14:00, Alexander Klimetschek <ak...@day.com> wrote:
> On Tue, Sep 29, 2009 at 13:09, David Wagener <Da...@awi.de> wrote:
>> But when I compare the performance of both solutions, the search module
>> based on “Bundle File-System” is faster than running search module on
>> “Bundle Database PM”.
>
> The (lucene) search index implementation is independent from the
> persistence manager. Search times should have the same speed if the
> content and the queries are the same.
>
> The only difference could be in the time it takes to index data, as
> that depends on the latency of the pm.
and the time to read the items of the result set from the PM.
regards
marcel
>> My Question: What about storing data in oracle? is there a possibility to
>> use oracle indexing rather than lucene's?
>
> Not easily. The search index implementation is made purely for JCR, to
> support property and full-text indexing plus all the special features
> as described in the spec. And Lucene, especially if the index is kept
> locally as now, is faster as an Oracle based index... at least I would
> guess so.
>
> Oracle itself offers a JCR binding (they were part of the standards
> committee), but only for their XML db AFAIK and I don't know if they
> support the full JCR search features.
>
> Regards,
> Alex
>
> --
> Alexander Klimetschek
> alexander.klimetschek@day.com
>