Mailing-List: contact users-help@jackrabbit.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@jackrabbit.apache.org
Received-SPF: pass (athena.apache.org: domain of aklimets@day.com designates
 207.126.148.182 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <22607975.post@talk.nabble.com>
References: <22587989.post@talk.nabble.com>
	 <c3ac3bad0903181441k5e648fccw583c9045509fbfb6@mail.gmail.com>
	 <22607975.post@talk.nabble.com>
Date: Sat, 21 Mar 2009 01:25:42 +0100
Message-ID: <c3ac3bad0903201725h6fdf38bercaf0672577a8195b@mail.gmail.com>
Subject: Re: Datastore(MySql)
From: Alexander Klimetschek <aklimets@day.com>
To: users@jackrabbit.apache.org
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: quoted-printable

On Thu, Mar 19, 2009 at 8:44 PM, Nichi <nichganjoo@yahoo.com> wrote:
> Say, I have two tables Experiment and ExperimentData that I want to store
> using Jackrabbit=92s content repository. =A0Experiment has around a thous=
and
> rows and ExperimentData tens of millions of rows. There is a one-to-many
> relation between Experiment and ExperimentData.
> My understanding is that each of these tables will be the node types and
> rows in these tables would be the nodes, columns being their properties. =
The
> reference property would be used to implement the one-to-many relation.
> Is my interpretation correct? Could there be a problem because of the lar=
ge
> size of the database ?

A "generic" mapping like tables =3D node types and rows =3D nodes would
imply you get flat hierarchies with millions of child nodes of one
"experiments" node. This is inefficient in Jackrabbit (we estimate the
max. recommended number of child nodes per node at 10K). And
completely misses the advantages of JCR modeling. You should rather
find a good hierarchy for your data. Almost always working is the date
of the item, creating subnodes like "2009/03/20/mydata1".

See also http://wiki.apache.org/jackrabbit/DavidsModel

Regards,
Alex

--=20
Alexander Klimetschek
alexander.klimetschek@day.com