jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jukka Zitting" <jukka.zitt...@gmail.com>
Subject Re: Clustering - basic question
Date Fri, 29 Aug 2008 13:49:31 GMT

On Fri, Aug 29, 2008 at 3:37 PM, AdamR <adam.ross@runbox.com> wrote:
> According to the Jackrabbit wiki, all cluster nodes must store their data in
> the same globally accessible location. So, from a Jackrabbit point of view
> there is still a single point of failure and availability is not improved.
> Obviously this single persistant store can be clustered itself to meet
> availablity requirements - but what value does clustering at the Jackrabbit
> level add?

The main benefit from the current clustering feature in Jackrabbit is
that you *can* access a single clustered backend store from multiple
Jackrabbit instances. Before clustering support was added, you could
not run two Jackrabbit servers pointed to the same backend store.

Also, the clustering feature makes sure that all caches and search
indexes remain coherent, that metadata changes like node type and
namespace registrations are automatically propagated across the
cluster, and that observation works correctly even when changes are
committed by other cluster nodes.

That said, you are right in that the current clustering feature is
somewhat incomplete in that it depends on external clustering
solutions. See the dev@ mailing list for some discussions about
potential new Jackrabbit persistence architectures that would feature
built-in clustering.


Jukka Zitting

View raw message