Mailing-List: contact users-help@jackrabbit.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@jackrabbit.apache.org
Received-SPF: pass (idunn.apache.osuosl.org: domain gmail.com designates
 64.233.184.226 as permitted sender)
DomainKey-Status: good
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
        s=beta; d=gmail.com;
        h=received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references;
        b=DlyGU7S10+8TaOUM8mvogrcfs5bAlx8+/E8q7F5UhLlEw4dsRhE3+yzOGl5lsVNmmFTULX+KScAB/Mihl6oawR+erg/0vn0mFbAcjAP/mJuQfhvTfVmCiIMVHNH9DNCDUGUqzVTucw0uyCsSykldUa9DgxPVgqC0rH7cjRP1h74=
Message-ID: <96ab3ced0609120203p55c488bbw3d06bc2e24ec5e81@mail.gmail.com>
Date: Tue, 12 Sep 2006 10:03:05 +0100
From: "Michael Neale" <michael.neale@gmail.com>
To: users@jackrabbit.apache.org
Subject: Re: Performance problem with references
In-Reply-To: <ee43dh$t8l$1@sea.gmane.org>
MIME-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_Part_26917_25214059.1158051785786"
References: <ee43dh$t8l$1@sea.gmane.org>

------=_Part_26917_25214059.1158051785786
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

I gather you are referring to saving the node that has all the references as
attributes?

ie:

NodeA ---> NodeX *

Where there are 100 000 NodeX instances. NodeX.getReferences() would then
return just one NodeA (in this case).
So when you save NodeA - you are saving all the references - and I would
expect that to be expensive. However, when you save NodeX - that should be
reasonably fast right?

If that is the case, could you try and "refactor" your repository (is that
the right word ?? ;)  so that there is a child node of NodeA which has all
the references, and if you are making a simple field change to nodeA (not
the references) just save it? (not its child).

Just some thoughts... (I am still learning myself, so this would help me to
know).

On 9/11/06, Christoph Kiehl <kiehl@subshell.com> wrote:
>
> Hi,
>
> I'm experiencing an extreme performance/storage problem with references.
> In my
> case I got node(document) that reference another node(state) by reference.
> This
> state node is referenced by about 100.000 document nodes (still growing to
> about
> 1.000.000). As NodeReferences get persisted on every change, there is a
> lot of
> data written to the database on each added reference.
> What is the best way to handle this number of references? Is there a
> better way
> than saving the state uuid as string property an hence give up referencial
> integrity?
>
> Cheers,
> Christoph
>
>

------=_Part_26917_25214059.1158051785786--