Hello,

On Mon, Aug 8, 2011 at 1:02 PM, Alexander Klimetschek
<[email protected]> wrote:
> JCR / Jackrabbit itself fits nicely into one of the many definitions of
> NoSQL, if you look at the unstructured data part of it.
> Using other NoSQL storages as Jackrabbit's internal persistence
> implementation (persistence manager or data store) is possible and some
> attempts have been made, but it doesn't bring much value, as the interesting
> feature set of JCR is implemented on top of that anyway. The only advantage
> would be if you already have & use NoSQL storage X and want to consolidate
> all data there.

another advantage could be that hbase scales out much further than a
plain database. Also, I am not sure if it could be used to remove the
single point of congestion of writing to the database in Jackrabbit,
although, I have to add I am on thin ice here, as I do not know much
about that part of Jackrabbit

Another part I would be very interested in, is if NoSQL would exactly
be efficient for the use cases Jackrabbit has. I doubt whether
fetching many nodes from, say a hbase backed storage, is actually fast
enough? Perhaps it is an attempt like storing a single Lucene index on
a cluster (like the meant to fail Lucandra attempt): It is a fruitless
effort, and will never be comparably fast to having the index locally.

Regards Ard

> Cheers,
> Alex
> On 07.08.11 10:57, "Dayo Egbetola" <[email protected]> wrote:
>

Reply via email to