Clustering

[Julien Poffet <ju...@gmail.com>]

Hi,

I was wondering if clustering can work without a shared DB. I mean it is
possible to use a File System Persistence Manager, which will use a shared
directory between the cluster? In addition the clusters node will of course
use a datastore.

The point is that read/write to my database decrease considerably  the
performances. So I was thinking that the a PM which use a shared directory
among the cluster might be faster... Am I wrong?

Cheers,
Julien

Re: Clustering

[Thomas Müller <th...@day.com>]

Hi,

The Jackrabbit file system persistence manager does not support
clustering. Anyway it is not recommended for production - see also
http://wiki.apache.org/jackrabbit/PersistenceManagerFAQ#Bundle_File-System_PM

There is a commercial option: Day CRX. The default persistence manager
of CRX, Tar PM, is file based and supports clustering. It is very
fast. I can recommend it :-) Well I'm one of the developers.

Regards,
Thomas




On Wed, Nov 25, 2009 at 4:21 PM, Julien Poffet <ju...@gmail.com> wrote:
> Hi,
>
> I was wondering if clustering can work without a shared DB. I mean it is
> possible to use a File System Persistence Manager, which will use a shared
> directory between the cluster? In addition the clusters node will of course
> use a datastore.
>
> The point is that read/write to my database decrease considerably  the
> performances. So I was thinking that the a PM which use a shared directory
> among the cluster might be faster... Am I wrong?
>
> Cheers,
> Julien
>