remotable jackrabbit by just wrapping with xfire/hessian?

[Phillip Rhodes <sp...@rhoderunner.com>]

It's a requirement of my application to be able to add,modify,edit and read jackrabbit content from multiple jvms and multiple hosts.  I see references to a lot of different technologies (SPI, clustering with a journal) to allow for this, but I don't see anything that I can count on to use in production in the very near future.

I am considering implementing a jackrabbit repository in a servlet container that is accessed via a set of web services (xfire or hessian).   Clients would perform searches, retrieve files, etc.. by calling the web services.  Only one jackrabbit repository would need to exist.  

I would appreciate your thoughts on this.  The different deployment configurations are confusing for me (since I am new to jackrabbit) and I don't want to go down a road that is wrong.

Thanks.
phillip

Re: remotable jackrabbit by just wrapping with xfire/hessian?

[Jukka Zitting <ju...@gmail.com>]

Hi,

On 4/25/07, Phillip Rhodes <sp...@rhoderunner.com> wrote:
> It's a requirement of my application to be able to add,modify,edit and read
> jackrabbit content from multiple jvms and multiple hosts.  I see references
> to a lot of different technologies (SPI, clustering with a journal) to allow for
> this, but I don't see anything that I can count on to use in production in the
> very near future.

The most stable (and easiest to set up) solution would probably be to
use the JCR-RMI layer for remote access. The jackrabbit-jcr-rmi
component has been in production use for as long as we've had official
Jackrabbit releases, and there have been few problems reported against
it.

See http://jackrabbit.apache.org/doc/components/jcr-rmi.html for
instructions on how to configure and use JCR-RMI.

The only problem with JCR-RMI is that it's not really designed for
best performance, as it essentially requires a separate remote call
for each node and property access. Perhaps you can get started with
JCR-RMI and for example look at the clustering feature if you need
better performance.

BR,

Jukka Zitting